High Temperature Parts

Home , Nimonic

AGARD-CP-317

SMaintenance in Service of

II- C)3

.3 L

DISTRIBUTION AND AVAILABILITY ON BACK COVER Best Available Copy AGARD-CP-317

NORTH ATLANTIC TREATY ORC iZATION

ADVISORY GROUP FOR AEROSPACE RESEARCH AND DEVELOPMENT

(ORGANISATION DU TRAITE DE L'ATLANTIQUE NORD)

AGARD Conference Proceedings No.317

MAINTENANCE IN SERVICE OF

HIGH-TEMPERATURE PARTS

Papers presented at the 53rd Meeting of the AGARD Structures and Materials Panel held in Noordwijkerhout, the Netherlands 27 September-2 October 1981. THE MISSION OF AGARD

Mhe mission of' AARI) is to bring together tile leading peloRalitie,, of the NATO nations in tile fields of Neiencv- and technology relating to aerospace for the following purpose-%

of scientific and technical infonnation: SExchanging

.Continuously stimulating advances in the aerospace sciences relevant to strengthening thie coniion defence posture.

lnpro'ing the co-operation among member nations in aerospace researclh and developmenrt:

Providing scientific and tccmt,-al advice and assistance to the North Atlantic Military Committee in the field of aerospace research and de'. elopnment

Rendering scicultit'ic and tc:hmeal aslIstancc. is requested. to oher NATI) bodies and to nmember nations in connection with leýsvearh and devclopnielt problelm, in tfie acro.pace field:

r!'tosiding assistance to inemhbter nations for the purpose of increasing their scie ntitiL anid tech1nical pot en tiall

Recollmllmlending effectisc ways for tile melmjllber nLlat.tIlo,tO Use their research and development capabilities Ifm the conut on benefit of the NAI 0 co0in t1LtN -

I'lhc highest authority wit hini A(.A RI) is the National Delegates Board cowisising of officially appointcd senior representati', fiom each member nation. heli mission of AGARD is carried out through the Panels which are composed of c.perts aIppointcd by tile National Delegates, tile Consultant and Exchange Programme ,nd the Aerospace Applications Studies Programme. The results of AGARI) \%ork arc reported io the member nations and the NATO Authorities thiough the AGARI) series of publications of which this is one.

Participation in AGARI) activities is by invitatonionly and is normally limited to citize.s of' the NATO nations.

The content of this publication has been reproduced directly from material supplied by A(;ARD or the authors.

ISBN )2-835-030?7-4

Printedr hi Teichnical Ihditing and Reproductutm ! td fliarford Ilise, 7 9 ('iarlottv St. Londm. II'P 1111)

1i PREFACE

Over the years the Structures and Materials Panel has been concerned with the problems of gas turbine engine materials. The Panel sponsored a co-operative testing programme to analyse and predict low cycle high temperature fatigue life and this was the topic for a Specialists' Meeting, namely:

"Characterisation of Low Cycle High Temperature Fatigue by the Strain Range Partitioning Method" (Aalborg 1978. AGARD CP-2431.

Another approach was the use of ceramics, discussed at a Specialists' Meeting entitled:

"Ceramics for Gas Turbine Applications" (Porz-Wahn (Cologne) 1979, AGARD CP-276).

While it was felt that these Meetings had achieved their purpose for their specialist audiences, there remained a gap to be bridged between the experience and information of the maint'yvance engineer and what thc materials specialist iii tLc re'earcl laboratory could , ffer. This Specialists' Meeting on the "Maintenance in Service of High Temperature Parts" wa: con- ceived as a two way exchange between maintenance experience and materials science.

It was felt that AGARD should be involved because all NATO countries faced common problems in the increasing costs of engine maintenance and the scarcity "f strategic materials. Many of the problem areas were likely to have a common base in relation to service experience and materials behaviour characteristics, so that an exchange of views should benefit both the research and maintenance communities. The collected proceedings and the discussion sessions endeavour to provide such an advantage to those involved in materials for gas turbine engines.

On behalf of the Structures and Materials Panel I would like to express my thanks to all authors, recorders. session chairmen and participants. In particular I appreciated the help of the members of the Sub-Committee on the Maintenance of High Temperature Parts when I btcame Chairman and that of Mr George C. Deutsch who, though retired, returned to conduct the final technical and discussion session of the Meeting.

I-..FANNER. ('hai.'nan - Sub-Committee on Maintenance of High Temperature Parts

iiix CONTLNTS ~

Page

PREFACE by ).A.Fanner iii

INTROI)UCTORY REMARKS by H.P. van Leewen

Reference

SESSION I -MAINTENANCE EXPERIENCE

MILITArPY MAINTENANCE POLICIES AND PROCEDURES FOR HIGH TEMPERATURE PARTS- Will they be adequate? by R.B.G. Hedgecotk I

ENGINE I)EPOI MAINTEN..ANCE REPAIR TELIINOLOGY by J.A.Snide and W,J.Schulz

NIAINI ENANCE PROBLEMS IN GAS IURBINE COMIPONENI S A 1 I IE ROYAL NAVAL AIRCRAFT YARD, FLEETLANDS' by F.J.Plumb 3

MAINTENANCE EXPERIENCE WITH CIVIL AFRO ENGINES by I.Ph.Stroobach 4

LIFE EXIENSION AND REPAIR I

ENGINE COMPONENT RETIREMENT FOR CAUSE, by J.A.Harris. C.GAnnis. M.C. van Wanderham and I).L.Sims

IDEFECTS ANI)THEIR EFFECT ON THE BEHAVIOUR O1: GAS TURBINE DISCS by R-11,ieal 6

RUCORDER*S REPORT - SESSION I by M.G.Cockcroft RI

SESSION 11 - LIFE EXTENSION AND) REPAIR It

A Ti'IANIUNM SILICON COATING FOR GAS TURBINE BLADES by G.ll.Marijnissen

INFLUENCE DES TRAITEMENTS D)E PROTECTION SUR LES PROPRIETES MECANIQUES I 'ES PIECES EN SUPERALLIAGE par i.M.Hauwer. C.Durefl IaR.Picboir

RECONDITIONNENIENT I)E PIECES FIXES DU TURBINE PAR BRASAGE DIFFUSION'. par Y.HI-norat et J.Lemgourgue%

I REJUVENATION OF USED TURBINE BLADES BY HOT I()STATIC PRESSING AND) REHEAT TREATMENT' by K.L.Clicung. C.C.Leach, K.P.Wiliet and A.K.Koul I0

HIP PROCESSING - POTENTIALS AND APPLICATIONS by W.J. van der Vet I I

RECORDER'S REPORT .- SESSION Ii by A.J.A.Ntom R2

iv Reference

SESSION III - LIFE EXTENSION AND REPAIR Ill

REGENERATION OF THE CREEP PROPERTIES OF A CAST Ni-C-BA•SE ALLOY' by K.R.Tipler 12

REPAIR ANiD REGENERATION (iF TURBINE BLADES, VANES AND DISCS by H.Huff and J.Worlmann 13

A NEW APPROACH TO THE WELDABILITY OF NICKEL-BASE, AS-CAST AND POWDER METALLURGY SUPERALLOYS by M.H.Haafkens and J.H.G Matthey 14

RECOIIERS REPORT - SESSION III by i.A.Snide R3

APPENDIX

COMMENTS ON THE MAINTENANCE iN SERVICE OF HIGH TEMPERATURE COMPONENTS IN AIRCRAFT JET ENGINES, by G.C.Deutsch A

ii ii INTROI)UCTORY REMARKS

by the

Meeting Chairman Dr-lr. lI.P. van Lecuwen National ierospace Laboratory-NLR P'.O. Ho., 15 3 '30O AD,)timneloord. Netherlands

The present Specialists Meetng on the Maintenance in Service of iligh-Tempeiatue Parts is not a routine Under- taking in the history of th. AGARD Structures and Materials Panel. Traditionally the 5\Mi' has concentrated on problem, related to the airframe rather than the engine.

Even at its first meeting in Ottawa. JUlne I1)955, when it discussed high-temperature problems, this was in relation to a proposed AGARlograph on I ligh-Temperaturtire flcts on -\ircraft Structures. At the 5th Panel Meeting in Oslo and Copenhagen. April May I1);. Materials for Use at Flevaied Temperatures were the main topic. but again the lecture' that were held had basically the airframe in mind: this in spite of the fact that considerable attention was paid te, titanium and its alloy.s.

Yet at certain intervals we have occupied ourselves with topics that were of interest to the engine :ommunity. "Sin,:e so many representatives of that conmmunity are present now., I would like to tell yot0 some more of our activities in that field. For y.our in'ormation I have ilso lirepared a list of [ngine Oriented Publication1S under the Auspices of the -G.ARI Structures and Materials Panel.

-he Summary Record of the 50t Pane! Meeting containt-d an Appendi\ called the S•ope of the ,ork o" the Panel. Ilcre we find formally listed as topic 3h: The properties of aircraft and Ltero-vngit: tarerial. both metallic and non- mnetallic, including synthetics. adhesives and plastics.

This resulted in the following papers being presented at the Technical Sessions on Fatigue at fhe occasion of the 6th Panel Meeting in Paris. November 1057: Fatigue of Structural Materials at Iligh lTemperatures. by Prof. B.J.1.azan. University of Minnesota. and Corrosion S•clic et Protection des Alliages Rsfractaires Ni-Cr 80. 20. by, M.Mathieu. ONF-RA.

At that meeting a Permanent Materials Committee w•as formed.

Ihis committee stimulated activities in the high temperature materials field, not so much with an eye on present day problems. but much more in relation to future developments.

Topics treated included: Cermets. 7th Panel Meeting. Paris. March/April 1058. Unconventional Metals (Be. Cr. Mo. Nb. Ta. Ti, W and V). first suggested at the 8th Panel Meeting. Paris, October 1958.

At the lth Panel Meeting. Paris. April 1959, the SMP me, in its re-organized or•rm. incorporaling a Structures GroLIp. a Materials Group and a Flutter Committee.

The Materials Group initiated an activ ity which was to continue for a considerable number of x ears anti shich was devoted to the four promninent refractory metals: Mo. Nb, I a and W and their coatings. Several a.spccts were considered: I. Alloy development ' Co tatinLts 3. Production development 4. Welding 5. Basic Research phase diagranms. diffusion studies,. impurity effects, new fabrication techniqucs. oxidation sittdiest, De.sign dat:.

Vi Various publications resulted. but a major achievement in my mind was a handbook on the impurity analysis of tile four refractorv mactal,, which was the result of a collaborative ,tttdy that lated for nmany years

Another topic that was pursued was that orgraphtte as a high trmplrmttrrr trmifttraO material.

Iligh temperature mechanical and creep testing were tle ,Ubjeclt of two other co.llaboratiL'0 eftorts.

At the 11th Panel Meeting. Paris. October 19(,4. it was for the first time lhat a topic wa, taken up. that waLot inmmediate interest to tile designers and the users ol jet engine,- this was Ihe Lieveloplienl and teICAIg of protectim- coatings for jet engine materials. Unfortunately this did not result in an A(,.ARI)-sponsorvd collaborative programme. Presumably this was becauis several potential participants did not feel their commercial Interest, sufficiently saft- guarded.

It is important of course that there is another AGAkD Panel especially devoted to engines and their problems. This is the Propulsion an.t Energetics Panel- In I NiO+the SM\ Was invited to contribute tTOa PITP meeting on engine materials. Since thtat time liaison has been maintaincd with the PI P and the two Pancl.; hale assisoted one another in soliciting speakers for their respective conference:,.

In dite years that followed attention was again gi%c n to the more basic a.spects su:h as the ihermo-phv; 'ical propertic, of materials.

The 2 5th Panel Meeting. D3ay ton. Ohio. April May I OW)nmarks a sudden growth In the Panel's interest In engine problems. I tigh Temperature ( orrosion %ka, selected as a topic wortlh of Panel it teition.

This resulted in a Directory of Organizations. Investigators and Programmes in I ligh-Tcr-perature ('orrosion Research completed in 1')71. in a highly succcssful Specialists' Meeting on Iligh lenipe•rature (orrosion w hich took place InI Copenhagen in thle spring of 1072 and in a Hlandbook onl Basic D~ata in I liigh-1eiritirature Corrosion completed it 11)'4.

Another engine-related topic ccnsidered by the Panel was Ihgh-TenIperature ..ow-4-.clc Fatigue. First a state-of- the-art report was prepared on testing techniques and ntethdxls of analysis. Then a Specialists' Meeting was held In the spring of N-)74.

In parallel to the tITLCF project an activity Was UndertaKeh concerning In-Situ Composites. This programme was run similarly to that on high-temlperatutre corrosion. First a directory was prepared listing workers and organizations active in research and development in this field. Then a Specialists' Meeting on In-Situ Composites was held also at the occasion of the SMP '974 Spring meeting in Washington D.C.

The topic oi' high-temperature design problems was held alive by an ad hoc group which, after li:-tening to a number of pilot papers. proposed another activity in the field of low-cycle fatigue. This was to comprise a collaborative testing programme with as one of its aims to evalutate the strain-range partitioning method for the analysis and prediction of low cycle fatigue behaviour of engine materials. Sixteen laboratories participated. equally distributed to both sides of the Atlantic. The results of the testing programme were discussed at a Specialists.\ Meeting that took place in Aalborg in the Spring of 1978.

Itt parallel to the Low-Cycle-Fatigue Programme alt activity was started in the field of ceramics. Speakers were invited to present pilot papers and in the Autumn of 1"70 in Cologne. a Specialists' Meeting was held on Ceramics for Turbine Engine Applications.

At the 48th Panel Meeting in Williamsburg Va. April I Q71).consideration was givet to a proposal to stimulate the dialogue between engine maintenance engineers and materials specialists.

At tile 49th Panel Meeting in C'ologne, October 1471). a proposal was made to prepare a Hiandbook on .. eroclasticity for Turboinaachi ties.

Discussions continued on the aspects of engine maintenance and these have led to the organisal ion of the Specialists' Meeting on Maintenance in Service of iligh-Femiperature Parts to which you have been invited to participate. As you will see various categories of papers will be presented.

In the first cateeory fall the papers that deal with the tis.-s- exsperience, both military and civil. with inspection. maintenance and repair of jet engines.

Another category discusses defects and criteria for engine life and cause for retirement.

A third category considers novel repair techniques and rejuvenation of parts that have been in service.

viil I-mi a !::.ir thI caTeeorv ix concerned ).%it protcI dI coatings. to b' Jpplied in the nianitfacturing tageI.but I .o at rcgulir IluterO..L during sets ice Mec.

\Ve liae ained at btringiTog tog1lihor people with strongly dietrenm backgroundS. bitt all in omni way insolsed m the prolqenl of prosid lg plesent day atxr, rztlt %ilh reliable engines. That .%ill confinrm tTobe rejiable for mnippreci-bhe length of lime.

"Bydoing so me Ilmas aimned not onyllV t ilie nations that produce engines, but also at thoe that 11z,\e to opernie and maintain engines produced elsewhere.

The engine environment is characterised by heat and presure. Maybe the dicuhsions during the Specialisti, Meeting will crete another form of heat and pressure. But then. it has been said that sufficient heal and pressure can turn carbon into diamond. I am not expecting that after this Meeting the floor will be tound strewn with diamonds. but I am convinced that the results achieved will be valuable for some timie lo comne- and I wish Ott all a pleasarnt and fruitful meeting.

[f-P. VAN L.FEUWNNT

ENGINE ORIENTED PUBLICATIONS UNDER THE AUSPICES OF THE AGARD STRUCTURES AND MATERIALS PANEL

.1GA.-I)RI Adrbi.dr.i Reporrs

Fitzer. FL. Thermophvsical properties of materials. AR. 1 l0•. I itzer. F. Thermophysical properties of solid material6 Project Seution 1A. Cooper:ative thermal expansion measurements up to 1000'( AR.S1. P971. "Fitzer. E. Thermophysical properties of solid materials. Project Section 1 B: Thermal expansion measurements front 1000°T to '(,00(-. AR.314. 1972. Thompson. E.R. Technical evaluation report on A(ARD .;pecialists" Meeting on directionally solidified in-situ composites- AR.76. l)74. Drapier. JM.. Hirschberg. M.l. Technical evaluation report of the Specialists' Meeting ott characterization of low,-cycle high-temperature fatigue by the strain-range partitioning method. AR. 130. 1970

-1GARD Reports

Lazan, B.a. Fatigue of structuraltnaterials at high temperatures. Report 1 ,, l;5". Mathieu. M. Corrosion seche et protection des alliages refractaires Ni-Cr 80 20. Report 159. 1157. Deutsch. G.. Meyer. A.J.. A review of the development ofcermets_ Report 195. 1958. Ault. G.M. Syre, R. Basic research on the refractor in,:tals molybdernum, niobium tantalum and luntgsten. Report 482. 1964. James. W.A. A state-of-the-art survey of methods being used for the evaluation of coatings for super alloys and refractory metals. Report 509. 1905. Coutsouradis. D.. Mullins, B.P. Cooperative programme on mechanical testing of refractory metals. Report 508. 196S. (cditori Gucer. D).. Capa, M. Effect of loading frequency on strain behaviour and dantage accumulation in lo%- c1c fatigue. Report 572. 1970. ICoulsouradis. D,. Cooperative creep testing programme. Report 581. 1971. Faurschou. D.K.

Jaffee. R-I. tediton NA (0 nationa! reports on high-tensperature corrosion of aerospace alloys. Report )V1 10472. i)rap~er. J.M. Ad ho-C group on low-cycle Itigh-t elperatutre fatigue. Status report. Repott 604. 1)2.

Fitzer. E.. Therntophysival properties ofsolid tnaterial_ Protect Section II - oopcratnc mncaure- mients on lieat trantsport phenomenalt of solid materials at high tentperature,. Report 0I,. 1973. Anonymus D)irectory of rescari:h ativitie ,on in-sitit corvposites. Repo'rt t(-0. 1 )73.

Iw Psolnl,.ý VI il'cdoo lI1C'1II1d il t 7dtcl

~n~R.!\t~' I hitjrtol 1!,; lx, ot 11:id*im. znolý I'd,~i ¾11(wih IjIiin ii rII'!lJ !Col. 1>w M3;.I, PI

NIilic r. I- wdo w W so"AWII Mfor l'!'noh "ila on ofik ohd11!ii:d. nc(, s N I~ MILITARY MAINTENANCE POLICIES AND PROCEDURES FOR HIGH-TEMPERATURE PARTS - WILL THEY BE ADEQUATL?

AUTHOR: Wg Cdr R B G Hec6%cock Air Eng 32(RAt UK Ministry of Defence Room 326, Old War Office Building Whitehall, London SWIA 2EU ULitel Kingdom

SUMMARY

Compared with Civil use, military sero-engines tend to be exposed to the effects of working closer to their design limits, of more rapid and frequent change of operating condition, and of more exacting environments. For high temperature parts this mode of use generates three areas of concern; estimating the life of critical parts, deterioration in material properties, and difficulties in inspection and repair. Our filing methods are already complicated by an economic desire to move away from average fleet lines to consider tbh life consumption of individual engines. Newer materials are aggravating this by the number of parameters, all varying in service, which now affect the life. There is nothing new in the maintenance engineer's ideal of materials which do not deteriorate and are simple to inspect and repair. However the conflict between this ideal and the demands for lighter weight and higher performance has probably never been greater. Will we be able to develop maintenance policies and procedures to maintain a sensible ialance?

1. INTRODUCTION From the beginning the physical properties of materials and particularly those of high-temperature materials have been one (f the main controlling factors in the development of gas turbine aero-engines for military use. In fact, from a viewpoint back in the year 1940, miracles have been achieved. Much has been written on the subject and this paper is not intended to add to that aspect. It is biased more towards the problems &ad challenges that the use of modern high-temperature materials bring to the Royal Air Force operator; living, as we all do these days, within strict socio-economic restraints. Material has to be considered in relation to its overall effect on the aero-engine as part of a weapon system. Moreover, the weapon system itself has to behave quite differently in peace and wartime conditions. The maintenance man is faced with problems of life estimation, with all its facets, inspection, repair and containment of costs; both direct and indirect. From all of these factors maintenance policies emerge. Each of these areas is touched on separately in this paper. Of course this is merely for editorial convenience because, in real life, they are inextricably connected in very complex relationships and what appears at the end is often a compromise between several aspects. 2. OPERATIONAl WEAPON SYSTEM EFFECTS 2.1 The Military Aero-Enaine. The military gas turbine is probably unique amongst aircraft equipment. Few other items operate so closely to the limit of physical properties of materials, has to undergo such intensive developmtent of reliability and life after it has entered service, has such a high proportion of critical and life-limited parts and is so frequently returned to the repair/overhaul organisation to be dismantled.

Then with each new aircraft type the demand for performance forces us towards engines that are smaller, lighter and use less fuel for a given thrust. If we compare the Rolls-Royce RB '99 for Tornado with an early Rolls Royce Avon of approximately equal dry thrust, advancing technology has achieved a significant reduction in the size of the basic turbo machinery. Even with its reheat/afterburner the RB 199 is no bigger than the basic Avon. To a large degree this change is due to the production of materials which will withstand higher turbine temperatures and to the design of improved cooling arrangements. Generally both are involved. 2.2 Peace-time and War-time Conditions. It is only during relatively short periods offrlig-ht that e-gines really consume their in-built life. Although aircraft, and hence engine performance is primarily specified for their war role, in practice peace-time training sorties can consume engine life more quickly per flying hour than war-time operations. Fig I shows where training sorties are condensed to eliminate transit flying and intensify training benefit. Perhaps in peacetime we should make do with lower thrusts and aenee improve the turbine environment. To some extent this cam be done. For example with some aircraft we use a flexible take-off thrust. The thrust used Is adjusted to take account of aircraft weigkht, weather and runway conditions. However many pilots feel, with some Justification, that the only good piece of runway is the portion that is left in front of you. For them. use of maximum thrust equates to jmatroved aircraft safety at a very critical point in the sortie. Of course a more mechanical solution can be adopted and we do consider for new aircraft the fitting of "combat switches". These would limit thrust normally but provide a facility in the cockpit to ottain maximum thrust for realistic training or in ar emergency.

2.3 Overall Maintenance Philosophy. Like many other aircraft operators, the RAY has tried many pilosophies of maintenance in the past. We are now firmly settled on a philosophy of On Condition Maintenance. This is a controversial phrase and many people define it in different ways. More importantly, they also understand it to mean different things. Many papers have been written and no doubt will be written on this subject alone. However here we mean that we try to get away from overhaul lives fixed for a given engine type and move towards lives that differ between individual engines and their components, and that depend on their physical condition, history of use, and whether their failure would actually hazard the aircraft. In practice the rules are applied to varying degrees on RAF engines in service depending on the type of usage monitoring equipment fitted to the aircraft concerned. As far as the turbine end of engines is concerned, our general rule is that because turbine discs are prone to low cycle fatigue (LCF) and because any failures might not be contained within the engine, lives based on engine uscge are established. The approach for turbine blades is rather less straightforward and depends upon the likely failure mode. Creep and fatigue can probably be covered, but high cycle fatigue presents problems. In general, because failed blades would stay within the engine we tend to impose no in-service life and just let the blades fail.

3. EGINE USE AND LIFE ESTIMATION. The subject of engine usage and life estimate is introduced at this stage because many people will be already aware of UK activities in this area from the detail in References I and 2. However a broad review is necessary because of its importance in relation to our other work.

Since our first tentative steps with an installation in a single Phantom in 1974, it has become increasingly apparent that total engine usage information was essential for two aspects; firstly to be fed back into aero-engine, and hence materials, design, development and testing. Secondly it provided for a more economic use, in-service, of the lives of engine discs obtained froe theoretical and rig test sources. In essence; the designer can now know how we actually use an engine as opposed to how we said we intended to. We can reduce the amount of unused safe life left on discs when we throw them away. 3.-1 Engine Usaae Monitoring System (EUMS) Mark 1. EUMS Ilk I is our basic equipment. It is used primarily to produce a fleet exchange rate between low cycle fatigue cycles and flying hours or sorties. More than 15,000 engine ground and flight hours have been accumulated on it. Twelve different types of aircraft are fitted with the system and seven more, Tornado, Sea Harrier, VC 10 tanker. Chinook, Puma, Jetatream and Sea King, are planned for future fitment. Normally a sample of 4 to 6 aircraft of each type has EUMS fitted.

Variations in engines, pilot operation, and sortie pattern are taken account of using statistical techniques and are written into a computer programme usec by the RAF. The following table, also used in Reference 2, shows the safe cycles per hour, on average, which would be expected ior -arious alternatives. In this case the figures happen to be for a compressor disc currently in service.

Safe average Monitoring Scheme cycles per hour LP HP

None 3.26 4-73

By role and side of aircraft 2.56 1.39 By role, side and sortie pattern code 2.42 1.1 By LCF counter 2.27 1.26 As a measure of the success of the system, the IF fleet exchang rate for the Adour in Jaguar was reduced, allowing time-e:pired Group A cnmponents to be reclaimed from quarantine at a cost saving of £1.3 aillion. in the same way EUMS data from the kegasus engine in Harrier is being used for life extension. An average increase of 20% o turbine diGca has been areed.

The RAP Red Arrow display team, now operating British Aerospace Hawk aircraft, present a special problem. EUMtS has shown that the outside aircraft in the formation can use about 13 times more LCF life than the formation leader on the same sortie. The need for recording on each Red Arrows aircraft will be covered later.

Not only does EU14S data provide ths engine designer with a knowledge of throttle movements and therefore cyclic life consumption on aircraft in service but the data can be used to predict engine behaviour in future aircraft. In order to simulate the mission profile for projected aircraft, a EUM1 fitted existing aircraft can deliberately reproduce tha mission. Where the new mission profile is very complex, EUMS sorties can be flown on several types of existing aircraft and spliced together- This exercise is also very valuable in providing for the differences between peace-time and war-time iperations at an early stage.

Currently the UK establishes disc life on 'life to first crack' principles. The RAF is very interested in any prugress on the latest materials towards a Fracture Mechanics approach where cracks were permitted up to a certain size; although we are conscious of enormous practical difficulties of managing this in complete engines and ensuring its cost-effectiveness. However, to obtain the maximum benefit, it would not be sensible to have to use "nominal" mission criteria based on the worst flying profile. Therefore, engine useage monitoring on each individual aircraft appears to be a necessary step in this case.

As well as EUMS fitted to aircraft, a number are in operation on test beds. Here experience is gained and LCF/hot end life consumption rates caa be quantified during the 150 hour Type Test. Moreover, the system can be used for more specific tests such as an assessment of the use of optical pyrometers to measure -irbine temperatuxes directly on military engines. The results can then be refs across to subsequent ones when the aircraft flies with the same modification.

The RAF is currently undertaking a trial using 12 Hawk aircraft fitted with EUMS primarily to explore the benefits of engine flight data recording and analysis "using a Ground Processor based at the flying station. Detailed fleet menagement is the area most likely to show results. However, the volume of data will be large and special arrangements have been made to relate it to engine ground performance testing and the physical state of the engine found during overhaul. Much will be added to the engine designers knowledge of the Adour engine itself.

In-service experience with EUME Mk I led to consideration of areas where its capability could be usefully increased and where extra features consistent with an overall engine life monitoring approach, could be added. The result was EUNS M 2.

3.2 EUMS Mark 2. Plessey Electronic Systems at Havant have developed EUNS Mk 2 to a UKinisty -- pecification. It is a microprocessor based system which includes a display suitable for either cockpit or equipment bay mounting giving parameter exceedance warnings and access to engine life data without having to use a ground replay equipment.

The system can accept 40 analogue parameters plus 16 spool speed inputs and 20 discrete ON/OFF signals. Up to 50 parameters can be checked fc: limit exceedances. LCF consumption is available for a number of critical components, both as a rate per flight and accumulated total. Data can be stored in either solid state memory or Quick Access Pacorder.

SUS ?"_ 2 is already flying in the Jaguar aircraft and is recording airframe fatigue as well .s principal engine life parameters. The system has now been installed in the Harrier.

The overall management of SUS Mk 2 and its data recovery follow closely the arrangements established for EU07 Mk 1.

3.3 low Cycle Fatiue Counter (LCFO). As a partner to EUS fitted to a sample number of aircraft, the £l07 provides the possibility of a monitor on each individual aircraft. The LCFC which is manufactured by Smiths Industries to a UK Ministry specification is a microprocessor based equipment which computes the LCF life consumed by 4 specific features in the engine rotating components. It notes rpm excursions and 'tonverts them into reference stress cycles. The RAF evaluation has taken place on aircraft already fitted with LVll so that the algorithas used in the LCF1 can be cross-checked. 14

Most of our results have been obtained on the Adour engine in Hawk and Jaguar. Its primary use so far has been on the Red Arrow display team aircraft where, as mentioned earlier, wide variations in LOF consumption can occur. At the moment the counter neglects thermal effects and the simple LOFC is clearly "restricted where thermal gradients or thermal fatigue are significant. A" it happens, it is safe to ignore thermal effects on a large number of RAP engines but this will be less possible in future. Three further approaches are being worked on. Firstly, the 1CFC is being developed to take account of turbine blade thermal fatigue and creep. Secondly, it may be possible to adjust the value of Ultimate Tensile Strength, used in the LCFC, to account for changes in material temperatures. Finally, FUMS Mk 2 could be fitted to each aircraft allowing more parameters to be measured. 3.4 Hot-end Monitoring. For the turbine area, in particular, the service engineer needs an on-oard hot end monitor that he can trust. Backing by a proven analytical basis is essential. The design process, including test programmes is fairly well established. The EUMS programme will be used to confirm life predictions. Rolls Royce and NGTE have developed creep and thermal fatigue algorithms which will be validated by phased removal of components and metallurgical examination. By ensuring that rs.asurements of compressor outlet temperature (turbine blade cooling air), turbine inlet temperature and shaft speed are available, the spare capacity built into the LCFC could be used to assess creep and thermal fatigue. This facility i:ill be available soon. However, a programme of EUMS correlation will be essential before the LCFC could become executive for these quantities. Even then both hot end equations have limitations because the failure mechanisms are not completely understood. However, a period of steady investigation should improve this situation. 3.5 Modular Engines. Before turning to the practical activities of inspection and repair, someUbroader mechanical constraints should be considered because they affect how and when these activities can be done. The RAF is firmly committed to modular engines because of the large cost savings from the reduced buy of complete spare engines. Moreover, engines can be recovered for use more quickly in the field. The newer engines have up to 17 modules. Most inspection and repair work is now done at flying stations with engines and modules only returning to an RAF Maintenance Unit or Industry when essential. This balance is the complete reverse of that employed in past years and, of course, the RAF still has large numbers of 'traditional' engines in use. The very flexibility of the modular engine and our in-service overhaul system also seriously compounds the management problem. For instance, we are progressing from the relatively simple task of managing a fleet of engines with the same life, through the stage with our Phantom and Nimrod engines which each have one or two different module lives to the multi-modular engine like the RB 199 example on a station operating 50 twin-engined aircraft; instead of 100 individual engine lives, we have 2,9002 It is for this reason that we have been forced to install a local station level ADP system based on the Texas Instruments 990/10 minicomputer. 3.6 Inspection. In an ideal world the maintenance engineer would like to be able to inspect all parts of the engine without dismantling anything and without removing it from the aircraft. Moreover, all defects should be so obvious that no inspector would miss them even at the end of a tiriag work shift. Real life does not allow such luxuries. To be fair, facilities for inspection have probably never been better. On the other hand there is more to look at; the number of blades is a prime example. Also the tendency towards more complicated fabrication reduces access to the areas to be inspected. 3.6.1 Inspection of the Assembled Engine. Here visual inspection still predominates. These days the engine is well provided with access ports fur borescopes. Even the problem of preventing access port plugs in the turbine area from seizing up seems to have been solved.

It has been clear for some time that the strain of lengthy unaided bore.;cope inspections greatly increases the risk that an inspector will miss a detect. Closed circuit television (CCTV) is an essential step both to ease the strain on the inspectors and also to allow a group to view and discuss a defect together and, if necessary, to take a photograph. Nevertheless black and white television can also be misleading and suggest spurious defects which turn out to be merely discolorations of the material. It is most likely that we will soon turn to more widespread use of colour television to counteract this problem. X-Ray techniques are sometimes used; either by X-Ray generation from the outside or by the introduction of a radio-active isotope into the internal parts of the engine. In general, the use of X-Rays is far from straightforward because the man' layers of fabrication cause a very confused and complicated X-Ray picture. Even so the technique has been valuable for some very specific applications.

AO.2 !nspection of the Dismantled Engine. 'Lhe most recent changes in this area for the RAP have"been introduced less by new techniques and more by where, in organisational terms, the inspection is done. As noted ear]ier the station level engineers now see more of the engine internals than at any time in the past. However, most module inspections at stations are limited to visual techni ues; both aided and unaided. Visiting non-destructive testing (NDT) teams supplement this. The RAF has developed, over the years, a fair capability for NDT techniques; spurred on by airframe structural needs. For engines and modules which return to Maintenance Unit level, the full range of NDT devices is available for use and most of our present effort is directed towards semi-automation of these techniqnms to increase productivity.

:. h_ r. On RAY flying stations very little repair of high temperature material~s •undertaken except by replacement of complete components such as turbine blades. However, removal of modules on stations exposes other modules to view. As a result, cracks in. say, the combustion chamber become known about. Whereas in the past if the engine was performinF satisfactorily, they remained invisible until the engine was finally dismantled. This new situation has increased the flow of questions to the materials designer about acceptable crack or damage limits. Here it is important that the advice given is sensible and that defects that we can live with are clearly identified. Wherever possible, the over-safe and expensive route of trying, to repair everything should be avoided.

Our modular policy introduces a further risk that, because the engine internals are exposed more often, they will even with our best control be exposed to minor and sometimes unnoticed surface damage. This is a fact of life for the future and the materials designer should bear it in mind. Although how he could possibly cater for such a random event is difficult to imasine.

Corrosion and erosion are always with us. However, they are areas where military operations can still catch out the maintenance engineer and indirectly the materials designer. 6ometimes our aircraft are eventually used in roles and environments never expected when they were first specified. A classic case is that of the Harrier used in Belize, w:here severe turbine disc corrosion/erosion was initiated by the unusual environment. At Maintenance Unit level the RAF share full engine repair with Industry. The cost of repair equipment required for modern materials limits its use to a few locations. Electron beam welding equipment, plasma spray and extensive heat treatment plant appear to be an inevitable price to pay for materials now coming into use.

;.8 Costs. Although the need to contain costs has driven most of what has been said aready, it is notoriously difficult to pin them down. In general the value of the aero-engines owned by the RAF is about £1,400 million and rising rapid]y with the introduction of the Rolls Royce RB 199. What is more, by the time an engine finally leaves the iervice, supporting it will have cost us at least 3 and probably more, times its purchase price, and that excludes the fuel it has burned. It is therefore usually easy to Justify work on more precise lifing and increasing the capability for inspection and repair. 4. CONCLU6IONS. Cost constraints and the need for increased engine availability has provoked a much more flexible approach to the estimation of the in-service life of engine parts. Therefore the UK will continue to operate and expand an engine useage monitoring programme, exploiting the strengths of the vahious equipments available, in order to refine component lives and move further towards an On Condition Maintenance Policy. At the same time the flow of data back to the designer should fill earlier gaps in the informatiun about military engine use. Good analytical foundations for predicted failure modes are needed early to identify the parameters to be measured and the algorithms to be used.

Techniques of inspection and repair in-service will, of necessity, evolve with future changes in materials and this need should not be overlooked in the search for materials bringing performance gains.

Although much of what has been considered applies equally outside of the high temperature materials area, this area does involve the shortest in-service lives and greatest difficulties of inspection and repair. Looked at positively, improvements in these materials therefore bring the greatest potential for general cost and operating benefits. ACIGOWLRGIPMTS ft. author wishes to acknowledge the co-operation of associates in Industry and also to his eollean in the Ministry of Defence for assistance ia the preparation of this paper..

The views expressed t- the paper are those of the author and do not necessarily reflect Ministry of Deferae policy.

REFERENCES

1. Hurry, MF and Holmes, M, Military Engine Usage Monitoring Developments in the UX, ADK Pape. 78-GT-65, April 1978.

2. Hurry, NP and Hedgecock, R30, Engine In-Flight Data Collection and Analysis in United Kingdom Aircraft, AGARD Paper AGARD-CP-293, January '1981.

0 0 4C c

LU -'

LU I .I

i 0 I e m0 ENGINE DEPOT MAINTENANCE REPAIR TECHNOLOGY

Dr James A.Snide Director of Graduate Materials Engineering Program University of Dayton College Park 300 Dayton, OH 45469. USA and Mr William J.Schulh Technical Manage: for AFLC Support AFWAL Materials Laboratory Wright-Patte.. 'n Air Force Base OH 45433, USA

The scope and mission of the two USAF engine Air Logistics Centers are described.

The various processes and organizational structure to identify repair technology requirements are discussed. Approaches to transition and implementation of new technology into a repair depot environment are described. Specific examples of technology developments described are: braze repair, laser metrology, electrophoretic, coatings, sputtered

MCrAIY overlay coating and inlet guide vane vibration damping.

INTRODUCTION:

The Air Force Logistic Command (AFLC) is the command which is responsible for the overhaul, repair, spare parts and supplies necessary to keep Air Force weapon systems combat ready. The responsibilities include aircraft, missiles and support equipment. As can be imagined, this is not a simple or small task. If the resources of AFLC were compared to those of U.S. corporations, AFLC would be ranked llth in the Fortune magazine list of the top 500 U.S. companies. In order to satisfy its mission, AFLC functions on both a contractual and organic or in-house basis. Five Air Logistics Centers (ALCs) have been established to provide the support to specific weapon systems or components. The primary workload assignments for each ALC are shown in Figure 1.

The organic repair functions of AFLC represent a stgnificant percentage of AFLC resources. The value of AFLC maintenance facilities and equipment exceeds $1.3 billion with covered floor space in excess of 16 million square feet. Thirty-five thousand people are employed for organic maintenance. In terms of aircraft and engines, these facilities perform maintenance and repair of about 50% of the aircraft and over 80% of engines overhauled in Fiscal Year (FY) 1980. Overhaul and repair of engines are accomplished at two Air Logistics Centers (ALCs), Oklahoma City ALC and San Antonio ALC. During FY 80 approximately 4,600 engines or engine modules, in the case of F-100, were overhauled by AFLC. Over 3,900 of these were accomplished on-site at these ALCs. Engine repair resources during FY 80 totaled $210 million. The distribution of the engine workload for FY 80, is shown in Figure 2.

For many years, Logistics Command was not viewed by the R and D community as a viable customer for new technology. The reasons for this are many and varied. For example, the cost of ownership of a weapon system was predominantly in development and acquisi- tion as opposed to operation and support, the logistics community was not requesting support, from the R and D community, and many of the problems encountered by AFLC did not lend themselves to waiting for long term solutions implied by R and D, and many laboratory personnel were not interested in applying themselves toward solving what they considered to be mundane problems. The close association between the Air Force Logistics command and the engineering community of the Air Force Systems Command has resulted in the identification of various opportunities for the translation of advanced materials and process technologies to the overhaul floor. These advanced technologies offer the potential for reduced costs lur- ing engine overhaul, Several examples of programs which have or are being conduct.Ad to enhance the repair of fan and turbine blades will be described. BRAZE REPAIR FOR TURBINE AIRFOILS:

Advanced turbine blades and vanes require the use of sophisticated air cooling techniques, costly nickel and cobalt base alloys, and extensive surface protective coatings. Because their operating environments cause various types of degeneration which ultimately lead to their removal and replacement, cost effectiveness of repair versus replacement must be considered in terms of overall lifa cycle management. The Air Force Wright Aeronautical Laboratories - Materials Laboratory (AFWAL/ML) conducted a Manufacturing Technology (MAYTECH) program with General Electric Corporation; (1) to establish cost effective repair techniques for conventionally cast airfoils. The objectives of the program were: 1) to select repair processes and airfoil types with generic application to ALC repair requirements, 2) to transition advanced process to manufacturing technology, and 3) to verify the repair procedure. Components selected for repair were the TF39 first and second stage turbine vanes and the first stages turbine blade. These are shown in Figure 3. Repairs performed on the Stage 1 HPT vanes incorporated cracked areas of the leading edge, trailing edge (con- cave surfaces) aft of the cooling holes and various inner and outer platform locations. In order to make a braze repair, the Fluoride-Ion Cleaning Process was developed in order to remove the oxide from the surface of the crack. A schematic of the cleaning process and its effectiveness in oxide removal are shown in Figure 4. Activated Diffusion Healing (ADH), which is a hybrid brazing process, was used to repair the cracks. The results of a typical ADH repair of a vane are shown in Figure 5. Representative components of each of the Stage I and Stage 2 vanes were subjected to 1,000 mission profile cycles in TF39 engine test vehicles, and inspected, and then evaluated to determine the degree of success for each repair. Repair feasibility for both components was satisfactorily demonstrated. The economics of the process for turbine stator components demonstrates the cost effectiveness of the repair concept in lieu of new parts replacement for total engine life cycle management. Additionally, the advanced techniques developed in this program using Fluoride-Ion cleaning and ADH exteiid the capability and cost savings afforded in turbine airfoil repair. METROLOGY OF COMPLEX AIRFOILS:

Measurement of dimensional characteristics of airfoil parts in both production and overhaul is primarily a manual operation. These measurements employ a variety of gage types. Gages vary from the very expensive optical comparator (Figure 6) to the inexpen- sive dial gage (Figure 7). Some gages are simple, others quite elaborate. All are subject to wear and require their own 4 ndividual calibration procedures. All gages require care in handling and some of the more delicate ones require frequent repair. The amount of training, skill and judgment necessary to use these gages varies consider- ably with the type of gage employed. Regardless of the dimensional gage used, alert and conscientious inspection is required by trained inspectors. Despite the diversity of gages and the heavy reliance on manual inspection, dimensional inspection today h satisfactory accuracy and provides adequate reliability. It is clear, however, that an automatic non-contacting inspection gage capable of measuring most dimensional characteristics of interest would: greatly simplify inspection operations: be cost efie:tive; consolidate a number of gages into one; minimize wear through the non-contacting feature; and improve overall inspection reliability by minimizing human involvement. The need for such a system is further emphasized if future needs are considered, e.g., the potential for increased airfoil parameter measurements and tighter tolerances that could result from the need for improved engine performance reliability. An automated system would address to future needs as well as provide many improvements for inspection of today's hardware. The features desired in an automatic gage establish the basic area for focus during its development. Rapid inspection speed is important for the automatic gage to be cost- effective. Accuracy is necessary to meet design requirements, and repeatability is important to provide consistent results. Non-contacting probes preclude scratching of the part and probe wear problems, and provide greater accessibility to part surfaces. Universality of application is desirable to minimize the need for many gages and their associated costs. Important ingredients of universality are ease of fixturing and flexibility of software to accommodate a variety of parts and inspection requirements. Automatic documentation must provide necessary information for proper part disposition in an easily understood format. Finally, the automatic gage must be compatible with the production and overhaul environment. All of these features must be considered and included, to the extent possible, in the design and construction of an automatic gage. The AFWAL/ML conducted a MANTECH program with General Electric Corporation (2) and Diffracto, Ltd. of Windsor, Ontario, Canada to develop and establish and validate a semi-automated, non-contacting gaging system to make accurate measurements. This program was initiated to improve this technology area by establishing a non-contacting, computer interfacial gage system that provides the capability of measuring many dimensional characteristics (e.g. airfoil contour, length, width, twist, lean, etc.) with one instru- ment while maintaining high accuracy, repeatability at reasonable throughput rates. The gage system was to be a demonstrator semi-automatic laser gage system capable of measuring representative dimensional characteristics of single turbine blades and single turbine vanes within a part size envelope approximately 12 inches long, 4 inches wide and 4 inches deep. The demonstrator system was to be capable of measuring ten representative dimensional characteristics of a test part. The test part for this program was the TF39, stage 2 trailing high pressure turbine blade. The demonstrator laser gaging system is a semi-automatic four axis system capable of measuring the ten representative dimensional characteristics of the test part. The demonstrator system (Figure BY consists of four primary modules: a) a laser sensor module; b) a part manipulation module; c) an output module and a computer module. The laser sensor and part manipulation modules comprising the demonstrator gage which inter- faces with the computer module is shown in Figure 9. The program essentially met the design criteria that the demonstrator system have an accuracy plus repeatability of 15% (i.e. +90 microinches) of the tightest tolerance to be measured at the 95% confidence level. The overall system met this requirement over an estimated 80% of the part size envelope and within 11% of the goal over the remaining 20% (i.e. the outer radial periphery). The major objectives of the program were achieved and a firm technological foundation has been established for a cost effective, reliable and flexible follow-on production model for airfoil part dimensional inspection.

ALUMINIDE PRODUCTION COATING PROCESS: Various aluminide type coatings are used to protect turbine hardware in Air Force engines. While these coatings are similar in composition and performance, their application methods vary so that -oating new and used parts may require separate installations unique to the processes. For example, the various steps in a pack process are shown in Figure 10. Cost would be reduced and logistics simplified if a single coating process were available at the Air Logistic Centers (ALC). The AFWAL/ML conducted a MANTECH program with Detroit Diesel Allison (DDA) Division of General Motors Corporation to provide a single process coating capabilty at a selected Air Logistic center. The Allison Electrophoretic Process (AEP) is to be employed for coating nine different alloys (nineteen components) in the Air Force engine inventory. It is expected that through the use of AEP the Air Logistics Centers could satisfy all their refurbishing needs with two basic bath chemistries - AEP 32 (Al-Cr-Mn) for nickel base and AEP 100 (Al-CrAl) for cobalt base superalloys - both being applied in a common installation. The two coatings proposed are licensed for use by the U.S. Government per contracts DAAJ-69-C-0412 and P33647-72-C-0007. Laboratory and engine testing are to be used to evaluate the suitability of AEP as a single coating process for Air Force use. The AEP Coating Process is based on the principle that migration is observed when an electrical potential is applied to two electrodes immersed in a dispersion of charged particles. An electrophoretic cell is shown in Figure 11. When the dispersion is properly formulated, the suspended particles will deposit upon one of the electrodes. Electrophoretic deposition differs from electroplating in that particles of iny compnsi- tion rather than ions are deposited. As compared with physical vapor deposition, electroplating spraying or dipping, the electrophoretic procedure is particularly suited to applying coatings to non-uniform geometries, because, as the coating deposits on external areas and edges they become insulated and the effective deposition shifts to uncoated areas. The AEP procedures utilize fine particles dispersed in a low viscosity polar liquid. Emperically determined coating weights are electrophoretically deposited upon turbine components. The coated articles are then subjected to thermal treatments in suitable environments (hydrogen and argon have been used successfully) for times necessary to produce substrate-coatings with the desired structure.

Facilities required for the AEP process are relatively simple and cost substantially less than for other conventional aluminide processes. A production facility has been operational at DDA for more than five years applying AEP 32 coating to Model TF 41 1st Stage Turbine Blades. Although this facility occupies only 250 square feet of floor space, it has a capacity for applying 'green coating" to 1200 blades per eight hour shift. A typical AEP coating facility and process steps are shown in Figure 12 and 13. An AEP aluminide coating can be tailored to meet a variety of specific turbine components requirements. The AEP 32 (Al-Cr-Mn) composition was developed to overcome thermal cracking problems encountered with the DDA pack cementation coating (Alpak) and at the same time hot,corrosion resistance was improved. The AEP 32 coating has shown excellent performance and applicability to a variety of superalloy substrates in previous AFWAL/ML sponsored programs (5,6). The resistance to hot corrosion, low and high cycle fatigue and stress rupture of various nickel base alloys coated with AEP 32 and alternative coatings are shown in Figures 14-17. Micrographs of the PWA 73, Alpak and AEP 32 coatings on two nickel base alloys are shown in Figures 18 and 19. Similarly AEP 100 (Al- CrAl) was developed to overcome Alpak coating thermal cracking and spalling on X40 and Mar-M509 turbine vanes, and is the prime candidate for improving turbine vane assemblies in Model T56 and DDA Industrial Gas Turbine Engines. The AEP 32 (Al-Cr-Mn) and AEP 100 (Al-CrAl) coatings have been successfully applied to a variety of turbine engine components st DDA for more than seven (7) years. This includes both solid and air cooled turbine blades, turbine nozzle vane assemblies ranginig from single to as many as six integral airfoils, intcgrally cast turbine wheel and blade assemblies and full-round integrally czst nozzle vane assemblies. Uniformity of finished coatings has been of demonstrated excellence; e.g. five years of production for the Model TF 41 ist Stage Turbine Blades with a specified coating thickness of 1.5 to 3.0 mils on gas path surfaces, airfoil, shroud, platform, and sta]", h.- shown a coating thickness uniformity over the coated surfaces of individual parts in 2.4

the range of 0.5 mil variation with no difficulty in meeting the overall specified thickness range. Similar success has been experienced with integral multiple vane and wheel assemblies.

As a part of the current MANTECH program, DDA is to install an AEP Coating Unit at the San Antonio Air Logistic Center (SAALC) to coat blades. The SAALC plans to coat initially the T56 blades and later the TF39 blades. Similar AEP coating facilities are planned for the Oklahoma Air Logistic Center that overhauls the TF41.

MCrAIY OVERLAY COATING PROCESS:

A class of coatings (the MCrAlY overlay type) has been developed that significantly increases the durability of turbine airfoils over presently used diffusion aluminide type coatings. The MCrAIY overlay coating is deposited on the airfoil surface and provides, independent of the base alloy composition, increased resistance to oxidation, hot corrosion, and fatigue. Because of the performance of tnese coatings and the fact that they can be readily tailored in composition and microstructure, they are finding increased usage in advanced military gas turbine engines. Such overlay coatings as NiCrAIY, CoCrAlY, and NiCoCrAlY, and improved compositions being developed, will provide the next generation of coatings for advanced military gas turbine engines. Increased durability and higher turbine airfoil operating temperatures are primary requisites for these future engines. Both of these needs translate to an absolute requirement that better-than-present protective coatings be integrated into the design and specification of the turbine blades and vanes. To meet the need, extensive coating developm'±nt efforts are being pursued. Many, if not most, of the efforts to develop improved overlay coatings are limited by the restricted capabilities of processing methods to apply tailored coating compositions a:td structures on an experimental and ultimately, on a production or overhaul basis. With the current electron beam vapor deposition (EB-PVD) process (used to apply today's MCrAlY coatings) it is difficult to deposit complex compositions containing low vapor pressure elements as Ta or Hf.

The AFWAL/ML conducted a MANTECH program with Pratt and Whitney Aircraft to establish and optimize an advanced overlay coating process which is suitable for application of MCrAlY type overlay coatings to high-pressure turbine býade and vane components during manufacturing and for recoating during overhaul for Air Force advanced gas turbine engines. The process is to be adaptable to both nickel and cobalt base superalloys with only minor process variations, such as adjustments or modifications of coating conposition and changes in coating morphology. An additional objective of the program is to provide a technique with equivalent or supeiior performance to commercially available vapor depý. -iona c•ating pcocesses with definable cost savings relative to advanced coating requirements.

Innovations in plasma-spray technology have resulted in significant improvements in the metallurgical quality of plasma-sprayed overlay coatings. However, laboratory and engine tests of NiCoCrAlY coatings within P&WA indicate that the best current technology plasma-spray performance is 60 or 70% that of equivalent electron beam coatings at temperatures and conditions defined for advanced gas turbine engines; therefore, sputtering was selected for further development and scale-up (7).

The sputter coating process is a vacuum coating technique in which inert gas io0s, typically argon, are accelerated from a plasma into the source, or target, mate.-ial to be deposited. On bombarding the target, the ions eject atoms and/or molecules ýrom a thin layer on the target surface to create a coating flux from the target surface. The part to be coated is held in the flux, and the ejected or sputtered material collected and recombined to form a solid coating representative of the target material composition. A schematic of the sputtering process is shown in Figure 2. The process shown is termed triode sputtering, due to the use of a thermionic emitter to increase the degree of ionization in the plasma. The cathode sputtering system is shown in Figure 21. A post-hollow target (cathode) geometry was used in this program for process optimization. A target configuration having two concentric, cylindrical targets. Figure 20, was selected as the best compromise of part volume-per-target area and deposition efficiency. With the parts to be coated, held, and rotated in the annulus between the two targets, coating material arrives at the part from both targets. Most of the coating flux material that misses the part deposits on the opposite target and is subsequently re- sputtered, producing a high material utilization efficiency.

A typical microstructure of MC-AlY sputtered coating deposited at low substrate temperature is shown in Figure 22. The apperance of the sputtered NiCoCrAlY coated blade after engine operations is shown in Figure 23. Because of the fundamental principles of the sputtering process, it has the fleribility to deposit coating composition with exceptional structural quality combined with the relative ease with which the process can be automated. reproduced and controlled, it makes sputtering an attractive production and overhaul coating technique. Some of the inherent advantages of sputtering are shown in Figure 24. This process will soon be working its way from manufacturing technology to the overhaul floor. VIBRATION DAMPING: The inlet guide vanes of the TF30 engine for the P-1ll Aircraft were experiencing premature failure in as little as fifty hours of operation. The a;alysis of the failures

m-' - t 2-5 revealed that the vanes were cracking due to fatigue as a result of the high stresses which developed at certain engine speeds. These excessive bending stresses in the inlet guide vanes were caused by the interaction of the incoming air with the inlet duct and the compressor. The initial solution required a redesign of the inlet duct and a retrofit "of the new inlet duct. This problem was analyzed by Dr. Jack Henderson of the AFWAL/ML. He developed a multiple layer energy dissipating adhesive and aluminum constraining layer which was adhesively bonded to the outside surface of the inlet guide vane. The selection of the adhesive, number of layers and surf. - treatment was complex because of the anti-icing requirement. A photograph of the damping wiap adhesively bonded to the inlet guide vane is shown in Figure 25. The effectiveness of the damping treatment can be seen in Figure 26. After engine and flight testing, the damping treatment was approved for application. This damping treatment is being installed on all TF30 engine inlet guide vanes during overhaul and is applied on new engines. Similar type damping treatments are being applied to the TF41 and TF33 engines. The success of this work has provided impetus for the development of high temperature polymeric damping materials and vitreous enamels. The application of damping treatment has demonstrated that this technology offers a solution to complex problems and an alternative to costly redesign. CONCLUDING REMARKS: In the past few years. considerable effort has been directed at reducing the cost-of- ownership of turbine engines by translating various advanced materials and process technologies to the overhaul floor. These technologies permit the repair of previously salvaged parts, enable parts to be inspected more reliably and more quickly, reduce the number of different coating facilities required and extend the life of overhauled parts. The various examples presented are just a sample of some of the efforts underway to address the problem of engine overhaul. Many opportunities exist in the engine overhaul environment for reduction of operation and support costs by the application of advanced materials and process technologies.

ACKNOWLEDGEMENTS: Special thanks to Mr. Sylvester Lee and Mr. Fred Miller of the AFWAL/ ML Wright-Patterson Air Force Base, Ohio for their help in obtaining information for this paper. REFERENCES: 1. J. A. Wein and W. R. Young, "Cost Effective Repair Techniques for Turbine Airfoils", AFML-TR-78-160, Vol. II, April 1979. 2. R. G. Alcock, R. A. Ekvail, R. S. rinter, D. S. Steel and w. H. Wright, Jr., "Manu- facturing Technology for Metrology of Complex Airfoil Shapes", AFWAL-TR-80-4062, December 1980.

3. B. S. Hockley, N. Wagner and B. Neilson, "Manufacturing Technology for Metrology of Complex Airfoil Shapes, Part II", F33615-77-C-5178, March 1980.

4. M. J. Barber, "Aluminide Production Coating Process", DDA Report EDR 10C519, December 1990. 5. K. H. Ryan, "Comparative Evaluation of Coating Alloys for Turbine Components of Advanced Gas Turbine Engines", AFML-TR-71-73, Vol Ii, January 1972.

6. Q. 0. Schockley, J. 0. Hodshire and T. Pacala, "Exploratory Development of Overhaul Coating Process for Gas Turbine Components", AFML-TR-78-84, July 1978. 7. R. J. Hecht, "Scale-Up of Advanced MCrAlY Coating Process", Interim Report, F33615- 78-C-5072, December 1978. COI"D, (AFLC) AIR FORCE LOGISTICS (MAJOR TRC ASSIGNmEnTS)

SACRAMENTO SAN ANTONIO WARNER ROBINS AGMC sm ALC AAC• ALlC cnhTY ACOLC oGDZN AC CXTY 2 4 2-AB.-5 ) F-15 Inertial Msmt Units/Platforms Nay. Inst.

Displacement F-Ill C-5A C-141 C-135 F-16 Gyros7copes

F-lOS C-13O E-3A CT-39 Electronic A-7D Weapons AGE Airmuflitioa FM-Ill Electrical/ Airborne/ Mech. AGE plectronics Oxygen Missile F-lO5 Nuclear/Com Life Support oxgnponents Sys Propellofs Components Comp. Landing Gear Engines (J79, Photo Equip Electrical 375,Engine TF30, (J57, j7 , T 3 ,F Components T56,- 100 TF39,)

TF33, TF41, Gyroscopes Fl01)(Except (except dis- R•a4 Air Turb) Training & Ground Elec Simulation (CEM) placement) Equip

Flight Con- Pneumatic Navigation (ex- trol Inst. Accessories Inst. cept IMUS/ Platforms)

Constant Pneumatic Hydraulics/ Access. Fluid Driven Speed Drive Rp" Air Tur- Accessories bine

(TRC) ASSIGNMENTS TECHNICAL REPAIR CENTER FIGURE 1. AFLC MAJOR Figure 2, FY 80 ENGINE PR~ODU3CTION (UNITS)

TMS ORGANIC CONTRACT INTERSERVICE TOTAL

J-5, 392 203 595 J-75 74 74 TF-30 438 438 TF-33 136 136 TF-41 257 257 T-58 30 30 T-64 12 12 T-56 364 4 368 J-79 682 682 TF-39 27 27 F-100-100 97 23 120 F-100 Modules 1,293 1,293 G/T-400 24 24 J-60 42 42 .1-65 12 12 J-69 105 i05 J-85 126 126 'T-53 2 2 T-7E 37 37 TF-34 21 21 IGS054 4 4 I0360D 87 87 0300-D 4 4 R1830 I I R2000 41 41 R2800 140 140

TOTAL 3,760 780 138 4,678 ______-______.1______Technical Approach I ~Select 1. Comp~onents_"e •e, - TurbineTF39 8trde

and Vanes

j-JFIGUR 3.E. F -O I ..

Technical Approach ,I III Oxides

i1 m: Establish SelectNew Component, Reai Processe~s

Received Crack Flouride Deoxidation Process IoL Turbine Vanes

After Fluorid. Cleaning

FICWRE 4. FLUORIDE CLEANINC OF OXIDE BEFORE B3RAZINiG Technical Approach

New ADH Repsired

Coionns Repair Cral

Activated Diffusion Healing (ADH) Of Turbine Vanes

A&Ft 4 CM" a

- -" 1 F-E: 5 ."CTIV1ATED DI-7FUSIONI OF-, ! UR INE VANES

FIGURE 6. OPTICAL COMPARATOR GAGE USED TO MEASURE TURBINE BLADES

I i 2,. -

FIGURE 7. DIAL INDICATOR GAGE USED TO.MEASURE TURB'NE BLADES

Demonstrator . System

i ~ ~~~Prts,Mantpulatlonand' L.,-,' S-No•'l ,.M 7:

FIGURE 8. LASER METROLOGY DEMONSTRATION SYSTEM

II--~.~-.. .-~-. Technical Approach

S eet AOH Repaired ComponentsCrck

Activated Diffusion Healing (ADH) Of Turbine Vanes

As A"NdA1 II~nG~.k& H 40

FIGURE 5. ACTIVATED DIFFUSION HEALING OF TURBINE VANES

FIGURE 6. OPTICAL COMPARATOR GAGE USED TO MEASURE TURBINE BLADES I 2-Il

. Demonstrator System Gage 1 "I,

FIUR-. LAE EROOYSSTMGG

-- m - + m m:

ALPAK COATING PROCESS

• - APPLV ALPAK

•.-_i•__ IGUR 10.ALPAK COATING PROCESS

LOAC PAC D T STF Ga T ILAST

APPYlAPA

= -- 4*AtiIM

FIGURE 10. ALPAS CETOLGY STECAES

|ii-A

I ' ' ~ ~IGUREI 0. ALA COTN PROCESS.. 1:+ l ...... ALLISON ELE-CT, :" ~ ' '-" f i:••"ES •

I I

AEP COATING FACILITY

"Br

COATINGl FACILITY FIGtURX, 12. TYPICAL ELTCTi•OPliROl•TlC

- '- - Atr. I." UA I INU r'ULtbb

LOAD,FIXTURE

APPLV AEP I MINVUTEAT RI'

UNLOPLOFIXTURE

DIFFUSE

FIGURF 13. ELECTROPLORETIC CO)ATINC PROCE3.-

WAI 14,OIU W T

]J~m700 !•, r0il-i AlI1 I l I l IfT" iT-i FS(UMRE 14. COMPAFHIOT I SON OF CORROSION RESIULTANCS oF AEP-32 COATED N1ICWEI,-OASE A[,IP01ii 14 cps

PI'(MIZEl 16 C(OMPAIPI ý-ON OPHC LI FI-: 01)>A- 3' "'ATI'l) N 1~IC1AV! AI I

Mil 4

00 4,t 70 attic 40 60

40 W 37 STRESS RUPTURE LIFE COMPARISON OF COATED RENE' 80 ALLOY "'c"

"p 31

36 38 40 42 44 46 44 50 51 1 , 460 20 + LOG HOURS) 1 10-3

F I L3 R E 1 COMPARI or' !,L*l`lTlZl- I, I ýFF-, 01' CODEP B- I ANU AEP- 3 2 CUP, VEU NEN-1 ii

FIGURE 18. MICROSTRUCTURES OF AlPAF AND AEII-32 COATEL, 713C 2-16

Ml~l

FIGURE 19. MICROSTRUCTORE OF PWA 73 AND AEP32 COATED B1900

-:-- IfA,'

iL•FTGURE 20. RICODFSTUCTUREW COFW 3ATNGD ARCS" EP2CO OAITEBI90 Mi=YOERA OT•_ST TIBN AROL

FTtYE 0.TRIODE SPUTTERINGCAIGPOE FOR DPSTN E~~YOVERLAYLCOATINGSTT1RNEAPIL TYPGI22 ICAL MICR~OSTRUCTU('TREGE Cv I 11IF.MCrAlYI 2 SPU lSITTEED CATIG D.W~EIEPOSITE~1EDATUF ENG19E, -()PkR4Tt0 OPUTTERED NICoCrAlY D BLADE

L ;all , PAý-ýY,

Typicaf Cas"q Platform

314X

ENGINE OPUFAINI'L) Si,ý-1-1111ý1:1)

-&PUTTE-RiN.G -PROCESSING ADýJWATAGE:tj Flexibility

Qualtly uniform"

RePraducilbilky Plus - Lowor Cost

ý4. -OMH AT1VANT:V.F.-" ý11' -!i;' FIGURE 2S. PROTOTYPE DAMPING WRAP ON TF30 INLET GUIDE VANE

INLET GUIDE VANE DAMPING WRAR

0 TF 30 -100 F/FIll F

, MIITIPLE LAYER ENERGY DISSIPATING ADHESIVES AND ALUMINUM CONSTRAINING 12 ~LAYERS

I 0 s14 MILLION PROJECTED SAVINGS

0 POTENTIAL FOR OTHER TF 30 ENGINES

o 12000 SPEEDIRPM)

VAN! FIGURE 26. COMPA'RISON OF DAMPED AND UNDAMPED INLET GUIDE UN MAINTENANCE PROBLEiS IN GAS TURBINE cOMPONENTS AT TH ROYAL NAVAL AIICPfl YAMwLnEnnx

F J PLUMB Senior Scientific Officer Naval Aircraft Materials Laboretory Royal Naval Aircraft Yard Fleetlands Gosport, Hants England P013 OAW

SUMMhARY

This paper deals, in a general sense, with the work of the engine repair facility at RNAY Fleetlands and describes the major problems found in the overhaul and repair of helicopter and marine gas turbines. Remedies for component reserviceability and developments to obtain longer service lives with a description of the techniques employed are discussed.

INTRODUCTION

The RN Aircraft Yard, 7leetlands, was established in 1940 for the overhaul of aircraft and engines for the Fleet Air Arm. Prom repairing of Swordfish and their internal combustion engines Fleetlands advanced to working on pure jet engines in 1954 and then increasingly with helicopter turbo-snaft engines. in 1967 Fleetlands became the tri-Service Helicopter Repair Centre and as the largest group within the Naval Aircraft Repair Organisation it is responsible for the repair and modification of helicopters, engines and components at various leveLr. Its ether tasks include the repair of hovercraft and the Tyne and Olympus Marine Gas Turbines as Change Units (MOTCUs) for the Fleet.

These engines can be life expired or may require specific work to be carried out such as repair of foreign object damage, rejection for engine health monitoring reasons, or part life refurbish of the hot end of a marine gas turbine. In addition sere engines may be received for defect investigation or as part of service trials.

Military engines generally will have a lower time between overhaul than those in civilian usage. This reduced time can be due to the effects of rapid temperature changes in the hot end of the engine and the frequent change of power output during service. Military aircraft and helicopters in particular operate at lower altitudes and therefore are much morn subject to the machinations of the environment than high flying Civilian aircraft. 7hese areas include operation in high humidity and salt laden atmospheres, making corrosion probably the greatest problem to overcome particularly within the Naval area. Sand pr-ovides in situ grit blesting and operation in tropical coastal areas produces all three, having dramatic effects on cold end components. At present, military aircraft have a low utilisation and this gives rise to other problems particularly that of corrosion before the engine would have otherwise completed its designated life. RN service engines are frequently compressor washed with fresh water for removal of salt which if done meticulously is very effective. If incorrectly carried out the effect can be worse since salt deposits would be washed into areas which operate at high temperature during service. Where deposits are sooty or oily, detergents are employed but there can be some adverse effects on some diffusion coatings.

The following paper gives a general description of engine over!,-e at DnAY with a description of particular problem areas found in hot end components and their repair. Obviously many of the schemes employed have been developed by the engine manufacturers but an increasing proportion have been initiated and developed within our own organisatiun as is particularly the case with the older engines. No reference will be made to the compressor system but it should be noted that this area probably produces the greatest proportion of problems found and particularly in relation to saline corrosion. Examples of hot end component failures are given and although these cannot be described as maintenance problems they do follow from problems in operation and 1 feel it is appropriate to raise them at this conference.

GENERAL OVERHAUL PROCEDURE

The aim of the repair process is to restore a degraded engine whose life has expired to a state that allows it to complete a further life. This procedure basically consists of 6 major operations: strip, clean, inspection, rectification/salvage, build and test. Of these no reference will be made to build and test.

Since all parts of a gas turbine are vulnerable tc dnmage or wear during operation each unit must be completely stripped. All the engine types are dismantled in a specific area by the same operators. Tne monitoring of components from each engine is controlled by a system of pallets; the components being segregated into specific groups in pallets thus allowing quick identification of parts and easy transportation over a system of roller conveyors. Once stripped and segregated the parts are dispatched to their own special treatment areas. The majority of parts will pass through a cleaning cycle, typically consist- ing of a paraffin' wash followed by immersion in one of several degreasing agents, a fresh water wash before drying by compressed air or in a warm air cabinet. In addition, items may require the removal of oxidation products. hOver the years the Yard has been faced with significant problems in the removal of high temperature prcteetlve paints from compressor blades without damaging the blaes. Progress has been made but it still remains a problem area. As great a problem Is related to the cleaning of hot end blades due to the tena- cious oxidation and combustion products. Hot caustic solutioaa are employed contalining vaious additives to aid conditioning of the oxide anýdits removal. Hot processin6 temperatures are in the region of 90 C. The hot chemical processing has been found still to be inadequate and the use of ultrasonic clea•ing associated with the hot chesical action produced beneficial results. More receznt experimen-ts using a vibratory polishing process allied with the hot chemicals appeare to have produced even better results.

Once cleaned and dried the components are in a r itable condition to be examined for defects. The components are surveyed visuslly or with binocular assistance as necessary followýng fluorescent or dye penetrant crack detection processing. Egine blades are processed in batches iinan automatic cleaning and crack detection plant using the fluorescent dye penetrant method.

Many of the initial stages of the turbine blades and nozzle guide vanes of our engines are pack alminised. The cleaning of these components is markedly more difficult than those without the pack aluminising mainly due to the roughened surface of the aluminised component providing a better key. The subsequent fluorescent dye penetrant inspection of aluminium blades can give cause for some concern. It would always be preferable to remove the coating in order to check the integrity of the blade. This is carried out on Marine Gas Turbine blades and nozzle guide vanes but not on our aero engines. The small aero engine blades would lose too much material from stripping and re-applying any alminised coating. Cracks have been detected in pack aluminised blades and there have been no spates of failures arising from this course of action.

Marine gas turbine blade and nozzle guide vanes due to the much poorer environment of saline atmosphere allied with lower quality fuel do suffer severely and suspect components are normally returned to the manufacturer for inspection and re-aluminising.

Following dye penetrant inspection the next stage is scrutiny of each engine component by a team of inspectors whose task is to reject those parts which fail to meet the standards laid down in the overhaul manual for that particular engine. Many rejected parts carn be restored to a serviceable condition ari are routed to various support facilities within the Yard such as the machine shop, plating shop, welding and thermal spraying groups etc.

The helicopter engines and marine gas turbine are dealt with separately due to the increased severity of envirorment in which the marine gas turbines operate. • "•LlCCT'ThR GAS TURBINES

"(a) Combusmion System

The combustion chambers of the older engines such as t:ne 'azelle, Nimbus and Onum-e are essentially manufactured from solid solution hardened alloys such as Nimonic 75 since they are readily produced in the form of sheet, possess good forming characteristics, and may be easily welded. h:owever the- do have only a limited high temperature strength and significant cracking is found it these older systems from stress raisers such as location slots, diffuser and cooling coles. Tirect weld repairs or weld repairs by replacement are simple and involve argon arc welding wit' ti-POCr filler wire having previously thoroughly cleaned the cracked areas. Unclean joints will produce porous welds and carn result in hot cracking. .,The arc gas is generally pure argon but Ar/H2 mixtures have been used to advantage to reduce porosity levels and improve weld penetration for a given power setting.

In areas close to the actual flame thermal 'atigue is a significant problem and there have been numerous failures in the older engines such as the early marks of the Gnome (Fig I). (1). These have been somewhat improved by the replacement of various sections of this paraicular system with Hastalloy X giving improved high temperature prope.'ties than that of Nimonic 7.

Carbon erosion used to be a problem in the Gnome but ham been e5sentially removed by modifications to the burner system, although we still see the occasional problem.

Tlhe modern combustion chambers are based on Nimonic C263 materials (Gem engine) a precipitation hardesned alloy. Our experience to date has been that every combustion chamber exhibits distortion and severe cracking between cooling/diffuser holes (Fig 2). Weld repairs following the approved scheme results in weld centre-line cracking due to the residual stress present within the component. IThe added presence of oxidation and combustion products does not help the situation. !he complete removaj of these stresses by solution treatment prior to weiding may solve this problem lut may result in further distorting the component due to its complex shape and thin section. This process is being examined along with the use of n high temperature braze repair techniose.

Fretting is a frequent occurrence at interconnectors and locating flanges. In the older engines these areas would typically be hard chromium plated and when fretted ce:ond acceptable limits repair would be sy re-hard chromium plating when possible or more frequently 1i-ywelding on a replacement part. Hard chromium plating is not the msgt suitable material for fretting resistance under these conditions since at temperatures above 450 C its hardness drops nr lyimuch that the thin co-r- ing is rnpidly worn through resulting in localised wear occurring i the softers parent rateriul, usually 'drhcnic 75. Re-plating then becomes impracticable. Plasma spraying using a nickel- chromium/chromium carbide composite is an effective rep•ir to saintsi:n designý dinmensio:.n with a coat- ig that maintains good hardness anld wear resistance at high temperatures .2l. fl-.ardrr s-r"nor- fretting resistant ceramic and tungsten carbide containing composites were not preferred 1 ecause the operat-ing temperat-;re was such that a degradation of these coatings would occur.

Cracking and erosion of the holes on the *;inmun diffuser shown in 'i,. is s frezu'ePnt pro e:.. A direct weld repair of tie holes is not possible h-uit replacement of th.e diffuser fro:.' can e reaiiu,' Carr -ed c-nt. t1s er "e rertrmed bye TI, o ee bat electron bmas weldin g ives a faster result particularly when processed in batches. 1raster reliability on weld oaazity is an added factor. It is anticipated that a sirilar tec:.niu:-:e can '-e em.loyed in. th.e Asatzou ' iJ ffuser wric' 5 ;ffers from toe same prm: len :.after a og lfe due to it.-'T'ite-a'e" - a ,Fig 4)_

Cb .ur-oi:e nec tio:.

,1irbine 'lades

In general we experience relatively few oroclems witn turnine clader wnere tnese 'lades are separate item.. .he gas wenerator tlacias are cast ilades of SE: Gnome and i s a:-. MAP-M-246 ,iem) and are unlifed. All are pack aluminised and invariasl-ý exhi it good corao- ion and erosion resistance. Tne blade tips however are frequently grornd to firal size during .uild and thus tnere is no protective coating at the dlade tips. This lack of coat2 ;g %as not resulted in any,, suhsequ ent damage oter t.an"a: in Nirbus engines and even then onlyin LIades exhibiting overheating.

In the case of the Nimbus engine we have been experiencing nigt levels of blade re.ectin,:.. A hagh percentage are rejected immediately for the burning of blade tips with a comparative number rejected for overheating ,tie 5(. In shiE iattew case sectioning of suspect b!ades is carried ou-t to con'firm the degree of overheatinz and to examine tor any, other defects, such as Creep damage, blade cracking and any creacvin- of tne anuninised laver. Where these defects do not occur the blades remaining of the stage =ay .be sib -ected to a full heat treatnen: as a salvage. Grain boundary oxidation of the blade tips in the direction of the blade root is b:eing round thus rendering the blades as scrap. The reasons for the overheating effects have not vet been established, but the modified Nimbus e-rinu has generated probiems in nozzle guide va.e distortion and cracking, the distortio:. of the shrouos frequently ;red-ucinetip rut, shown i% Fig 6 (5).

(2) urbine 'iscs and Bliscs

Tnese components are lifed and generally will fulfil their Jesigned lives before scrapping. Corrosion and overheating problems are very low in gas ienerator discs. However, certain components. notably- the Gnome power turbine disc and shaft suffer very badly from corrosion pitting on the disc faces, fir tree rcots and in the shaft bore (_ig -. Irials in the use of 0 the electrophoretic paint, £L19 . on Gnome hovercraft engines produced excellent results in that .o further corrosion pitting or defects developing from existing pits were found. -;owever, t.is potential solution having been found, material modifications fron KE(44E to A2P6 have been insti- gated and this is expected to resolve the problem.

The blisca (turbine blade and disc manufactured from th'e solid) are only found in the Nimbus (pre-N,od 719) an;d Astaznu ITIN engines. We still see occasional failures in the 2-nd stage turbine of the "imbus (pre-Mod 219) due to fatigue failure initially in the trailing edge of the blade. All of the Nimbus blades are pack aluminised ard give satisfactory performance from a high temperature corrosion/oxidation vi.ewpoint.

As yet little can be said on our experience with the Aztazou IIIN as we have only Just started the overhaul of this engine system.

(3) Nozzle Guide Vanes

The nozzle guide vanes are manufactured as a casting or more generally as a fabrication with vanes welded or brazed into the shrouds depending uponr its operating temperature. The 1st stage nozzle guide vane is manufactured from the higher temperature super alloys and would be pack aluminised, the following cooler stages generally of _imonic 75 fabrication with or without aluminised coatings.

Or. every engine that we overhaul cracking is always present, usually in the blade/shroud 'oints and blade trailing edges. Dependent upon the position, size and morphology of the cracks tney can. be tolerated, examples of which are shown in Figs 8 and 9. roe cracks are chiefly due to thermal effects such as thermal fatigue cracks on trailing edges or insufficient strength at brazed joints, or due to engine and aircraft vibration. Erosion and impact damage does occur b-it in ceneral this is not too serious.

Repair work may be conveniently split into those components which are al.uminised high temperature super alloys and those of the cooler non-aluminised Nimonic 75 components. Until recently 5NA'Yhas been restricted to Just carr:'ing out argon arc weld of cracks in the parent material of %75 components. Any braze repair work would have been returned to the manufacturer. Any such vacuum brazing repair was restricted solely to cracks in the braze. Where cracks branched into the parent material these components would be scrapped.

fleetlands has recently been carrying out its own programie on the use of high temperature braze repair using Ni-based braze materials. Pre-cleaning of the component is most important in order to remove combustion end oxidation products within the crack. This can be accomplished using the standard cleaning methods such as immersion in hot alkaline solutions, acia pickling or grit blasting followed by a hige. temperature anneal in dry hydrogen or high vacuum. This is then followed bý vacuum brazing using Ni-baaed alloys and a longer cycle time than normally employed in etanuard brazing practice. T-e aim being to carry out a diffusion brazing procedure in order to reduce and hopelully remove centre lire phases which w¢ ld be detrimental to the Joint mechsnical properties. The 4oint clearances of cracks vary from wide at the mouth to zero at tne tip and usually there are multiple cracks at varying orientations. This obviously creates problems initiallý- on cleaning the crack root but also on selection of the 3-4

brn•,e materials employed. The free flowing brazes required to pernetrate to the r, c , crack may not be suiteble for the crack mouth. Various combinations of brae- mixturec snI fier materials have been employed to overc me this and also to be compatibýle with the braze in,. rial used in manmfacture. A number of no7.'I,- rude vanes have been r-epairod usinv ?h, above technique to repair cracks and foreign object damage az.owt.iz. ýig it interesting to observe their condltion thrnough :r,.n-irmalsrvice.

With regard to the higher order super-alloys, ew Nimoni,- 9C and P26!, whi .n r amounts of Ti and Al, cieaning of the components becomes inf.reaeinly difflo lt as the. 1ve'n of Ti end Al increase since these elements are essent25ialy. •.mpToved t,, in -:e e hl,' t-Cpera- ture oxidation resistant 2 as well au improvs g the high temperature merincira] slra ni! . through precipitation of sub-microscopic JY' Luboids of Ni' Al, T i 'J . ertainlv -, va'1-M clearing operation will not be sufficient to clean the cracks and even ea', remely .ry I.'Jr-_ý.-. may be insufficient. The use of a more reactive reducoJng epce ya well be thle;inswer. Our work on this is contfnuing.

(c- Exhaust Caseings

The gas temperature in this area of a turbo-shaft engine tends to be relatively low and stc nsc as iS/8 austenitin stainless steel or Nimonic 75 are employed. The casingFs are of complex shape to deflect gases away from the aircraft fuselage and rotor w.h'ch when allied to the thin gauge materials used inevitably produce distortion and crooking. We have leveloped over the years various wI long and patching schemes other than those of the manufacturers for Weld repair and replacement of d-fc- ive parts. Frettage iL also a significant problem on flangec and we are am•n devlosir.• car owr. schemes to combat this either by replacement or by the use of plasma spraying for beilding 'Ipand providing frettage resistant coatings.

MAP'NE GAS TURBINES - TiNE AND OLYMPUS

a) Combustion Section

A major source of problems from the earliest days of the marine gas turbines have. been related to the fuel and combustion system r4). in both engines but particularly the Olympus the RN service requirements for smokte levels were never achieved using t••e standard multiflare can :Olympus) with th. original burner. Extensive modifications resulted in greatly improved smoke levels but at the cost of increased complexity. This in turn resulted in problems for }TCU interchangeability when matching spare GTCUs with the progress of module specifications. Combustion can life has been greitly improved from 1000 hours for the original to 2500 hours with the developed phase I cnan, whith when combined with the improved '&X burner design gives generally satisfactory performance.

Typically problems related to the czn are of cracking in the swirler vanes which may be repaired by high temperature vacuum brazing, as are the occasional cracks at diffuser holes. The inlet and outlet flanges are coated with a detonation sprayed tungsten carbide/cobalt composite powder and give sRtisfactory performance, although a plasma spray coating using the same composite powder may be a satisfactory alternative at lower cost and quicker turn round times. Until this summer RNAY had not received the uprated version of the Tyne RMiA, the RMSC.

Typical problsms of the early RMIA engines would include distortion of the Nimonic 75 combustion chamber, fretting at hard chromium plated interconnectors acidthe serious build up of carbon within the chamber which, when released, resulted in severe carbon erosion of downstream components. lmpro,:uments in burner design has greatly reduced this problem and given a resultent increase in engina life.

The uprated RMlC produces higher temperatures resulting in greater distortion and cracking of the combtŽstion chambers (4). Material modifications to the stronger C263 alloy failed to cure this problem. Subsequent developments by the manufacturer using Nimonic 86 with trermal barrier coatings are proving most successful to such an extent that engine lives of 5000 hours are now expected.

Olympus burners suffer from severe corrosion in the barrel area. The component is a 0.4%1i item forging with a 1O-1,ym nickel plated coating for corrosion protection. When exposed to sulphurous stmosphre (the sulphur mainly originating from the Dieso fuel) free sulphuric acid is produced. This, together with the fine salt spray found in marine engines produces localised pitting corrosior, followed by corrosion beneath the nickel loyer then subsequent severe attack. A weld repair scheme developed at Ileetlands salvages these components and the sesof ceramic barrier coating may well significantly reduce this problem.

(b) Turbine Section

As stated above, carbon erosion was a major problem on the early Tyne RM1A engines as shown in Figures 11 and 12. Carbon ýrosion still occurs but is not now considered to be a major problem.

With regard to hot end corrosion the literature is full of information of turbine blade deterioration under severe chloride ingestion and in the presence of sulphur, usually from the fuel. Since this is a major field of its own it is not proposed to discuss the mechanisms here. 'Ir to the present time only pack aluminised coatings have bees employed in RN service othrr than int ervie trials. Recent changes to the use of platinum aluminide coatings are expected to give longer 3-5

service lives with a less brittle coating. Corrosion effects are markedly temperature dependent - hot end corrosion is no different. There certainly appears to be 2 forms of corrosion taking place (sh,ow: in Fiss 13 and 14), each in its own temperature regime, and no coating yet developed has provided adequate protection against both forms. The implications from reported research work carried out in this field are the use of nigh chromium and intermediate aluminium levels in coating and hisn Ti/Al ratios in the parent material. Ie -ole of cop incs from the standard pack aluminising, through precious metal variants of alrmin- ising 8 , 1he n-re recent M Cr ALY coatings and their applications by diffusion or overlay techniques is very -ae an . 'tere is no doubt that this is an important area for long term development. This is especially so as longer service lives of gas turbine systems are demanded.

"NON-DETRUCTIVE TESTING

'7he standard techniques are employed in the examination of hot components with generally visual, dimensional and dye penetrant examinations with judgesents based on accumulated experience backed up by resident engine manufacturers representatives. As stated above, however, the one group of components which does give some cause lor concern e 'e pack aluminised blades due to the difficulty in cleaning the components without damaging the aluiminised costing. Cracks have been detected in such blades and as there has been no spate of failures we are reasonably satisfied although not complacent.

One further technique that we have recently started specifically on turbines of the Astazou TITN involves replicating specific areas of the component. These areas are mechanically and electrolaticall; polished to remove oxidation and combustion products and leave a highly polished surface. A plastic replica is taken of the surfaces and is then examined between glass slides under a microscope for defects which may not be found by dye penetrant techniques. These would include fine impact damage, corrosion, surging, creep and fatigue cracking. Mechanical damage on the curvic coupling will also cause grain boundary defects, but this is usually coarse enough to be detected by dye penetrant techniques. A certain amount of rework is allowable and this would be followed by a repeat replication of the area to ensure complete removal of the defects. The process is time consuming and requires operator experience but can show up significantly more detail than otherwise would have been obtained and allows expensive components to be re-used whereas they may have been scrapped.

Where components have been salvaged by welding, brazing or thermal spraying techniques there is obviously a need tu know that the work has been satisfactorily carried out. Such components would be subjected to dye penetrant examination and also X-radiography when required and where it is practicable. All welders are required to produce frequent test pieces which are subjected to critical examination, and if not satisfactory can result in the welder losing his "licence" to continue.

The testing of thermally sprayed coatings is much more difficult since the only real test is cne involving destructior of the component. The use of teat pieces is again required and it is essential that these be carried out under the same spraying conditions and those of component for them to be meaning- ful. This is particularly important where abradable seals, such as nickel/graphite, and boron nitride/ "cement powders, are employed. Too soft a coating will result in excessive coating wear; too hard a coating will result in blade tip wear or chipping of the coating resulting in impact damage downstream. Metal carbide containing coatings (eg WC, Cr2C 3 ) again would be tested to ensure proper fusion, distribution and morphology of the carbides, porosity levels and obviously for a clean costing/substrate interface by metallographic examination.

Bond coats of Ni-Al are normally subjectcd only to an occasional check to ensure quality of the -oating since these materials offer a wide tolerance in operating conditions during spraying.

Wherever possible the process should be automated to give consistency of spraying conditions between components.

Until recently there was no non-destructive test of the coating other than visually. Franke and de Gee (5) and subsequently Reiter (6) have developed ultrasonic methods to establish areas of poor contact between the coating and its substrate and within the coating itself and although this cannot be related to adhesive strength must give confidence in the technique.

CONCLUSION

In conclusion to what is a somewhat general paper RNAY Fleetlands experiences specific problems in each of the engine systems it is required to overhaul and these tend to be highlighted in component short- ages of particular engines, several of which are over 20 years old and still have to be kept in service. If specific areas must be highlighted these would include:

(a) Cleaning of turbine blades and especially coated blades prior to inspection.

(b) T.e problems associated with the repair of aero engine nozzle guide vanes and to a lesser extent, c1combustion chambers, which are individually expensive and almost always in short supply. Research carried out has produced successful results in the repair of the Mimonic 75 based NGVs but as the higher order Nimonic alloy assemblies certainly require more work.

Like any establishment working with highly complex pieces of machinery we can be subjected to the whims of supply and it is hoped that this paper zives some indication of our problem areas in the field of this conference and developments that we have adoptid or developed cArse)ves in order to overcome those difficulties. 3-6

REFFRmNtS

1980. (1) RNAY Fleetlands' Investigation Peport No 3574, by Plasma CoMbutZ.oL Chaw. or Inter-Conlector Sleeves 4(a F J PlUmb, "The Slv"e0 of Ge-sle E&gine no 6O-o02, 1980. Spraying', NAMLTechnical Memorandum Report No 345B3, 1960. (3) RNAY Flettlands' Investigation No 1, of Naval Engineering, Vol Ž6, Gas Turbines - A Beview", Jiornal (4' Cdr J M Kingslasd, RN, "Marine December 1980. London 1978. Surface Coating Technology", A W J,Paper 8. "Advances in (5) Franke, W and de Gee, of Bath, Private Communication. (C) Reiter, H,University

ACKNOWLEDGEMNS and the Department, RNAY Fleetlandf the help of the Engineering The author wishes to acknowledge J R Smith (NAML). advice of Dr K G Kent and Mr

represent the official view This paper expresses the opinions of the Author and does not necessarily

of the Ministry of Defence.

1961. bondon, United Kingdom 0 Controller, hISO 3.7

i.a lu re in ,irome co u tiot chaa ma er at t ae r'.age 1 nozze guide van.• locatJ]:•

arm

iii'uK.re' 2. <'rncktirg a• ah ilie " a a;nt of ai~:en cOc a;[•£! c~ m~ a *;, [I nr. : r if , r *

Ilk * n. a-, C4r

000

* 0

* -

i- ,e

,'a J*- .a a' !1-|

Figure 5. fverheating of Stage I tirbire bleder7 of týe Nirpbs Mpotrood -1%% enrio.e :Lwr the 'Llued' ppearance ii toe upper :':aif of the aerofoil.

-ig re C,. -j.jrL ie tip rub o:; the Nimbus Stage 2 r o7rle r'i'le vane rcr3 r~i: ýk of I e. -. r S.r. . .L-O

a.sha. .

Figure Corrosion pitii :) ta disc "ace .rl fir tree rt.ot:.e:o-leT.'

, xp

V---igureM?. orilhero ionrAlrdshaft, eltiy : of•era l' face•,ifr .•e '•t f t. •:eT wIV •.;. r .

I ..

!/, " -3 -1,

Figure 9. Cracking adjacent to the outer shroud braze in the Nisbus Stage 2 nozzle guide vane.

Figure 10. High temperature braze repair of cracks in the braze, guide vanes and foreign object damage of the component shown in Figure 9. 3-12

vanes at locating shrouls (left), Carbo;n aronion of Marine Tyne engine nozzle guide 'igure 11 as compared to a blade in good condition aerofoil face and bending edge (top right) (bottoa•, right).

jo,-

3OEK i 1111turbine blades, at 2300 hours life (right) and Figure 12. Carbon. erosion of Marine Tyne

11ours life (left). 4;

.,a

lgd

Air

V.4) 4 ______f' ilI o i J.-'

'- '

C, :

• I I . -. 10.') • ~Qr

-A6rI

:•___ PHigh Temperature Sulphidation .• 4)n

4;.- flighTempratueSuphid0io4

iC iC

i4 4-1

MAINTENANCE EXPERIENCE WITH CIVIL AERO ENGINES by ing. J.Ph. troobach Propulsion Systems Engineer Central Eng. Dept. SFL/CG KLM Royal Dutch Airlines P.O. Box 7700 1117 ZL Schiphol Oost The Netherlands

SUMMAijY

In civil airline operation the maintenance and fuel costs are a major item of concern and subject to close control in order to survive the hard competition between the airlines. Preventive maintenance of gasturbine engine High Temperature parts is mostly limited to visual examination and/or performance EGT trend shift triggers, aiming at avoiding extensive secendary danau.7 The trends in maintenance concept developments are reviewed, indicating the constant activities to optimize the maintenance cost of tihe propulsion system. As a result of the escalating trends in material and fuel prices, the presently applied maintenance concepts require a more sophisticated condition contro! in order to comply with the need to find the optimum operating time of each indivilual engine. With the introduction of the new generation of civil aircraft (Airbus A311,) a mutual goal between the engine manufacturer and the airline has been defined to develop mathematical program; based on actual recorded conditions, in order to control the behaviour of the engine aiming at an optimum use of the propulsion system.

INTi-ODUCTION For a civil operator like KLM it is of a prime objective to sell passenger and freight transportation in a fast and convenient way, competitive with other means of transportation but also economically justified and aiming at a financial positive end result. From this point of view, maintenance and fuel to be used is considered to be a negative contribution to the ratio of cost to revenue. The aircraft propulsion systems are, because of deterioration and fuel burn reasons, a major contributor to the total utilization cost of the airplane and thus subject to contizr.:cus iaprovementc in order to obtaii, the 6ighest propulsion t'rust for the lowest fuel burn and weight Penalty. Using gasturbine engines for aircraft propulsion a balanced compromise has to he derived between two extremes: - Based on toe Brayton cycle a high turbine inl~t gas temperature is a prime contributor to overall engine efficiency, however this has a direct impact on material cost and durability of the engine. - Operating with a low turbine inlet gas temperature allows less advances materials to be used, however the resulting efficiency loss has also a large impact on operating costs.

The practical side of the problem is that the past design efforts to build more powerful engines are superseded by the design aim to have the most efficient engine againut the lowest utilization and maintenance cost.

MATERIAL AND FUEL BURN COSTS Both material and fuel costs have escalated substantially in the past seven years, resulting in proportionally higher operating costs.

Materials: Cobalt +700% Tantalum +800% Titanium +400% Chromium +500% Fuel : Jet Al +700%

The composite price index in the same time perio! increased by only + 100%. Future trend developments are hard to forecast, e.g. a fuel price increase between 12% and 20% a year may be expected. Nevertheless the cause of the price increase is shortage of the major materials on the international market, and this condition only tends to get worse. The shortage on the market led to considerations as to which strategy has to be followed. Options are: - scrap parts reclamation for recycling. - selected use of materials development of alternatives. - expansion of exploration. There is not yet an active program developed for recycling of hot section parts, because. it is not yet clear who should initiate this program. The develonrment of alternativws might be the best long term solution. 4ý2

The direct and related cost increase 4 passed onto the final users of the maturia.s.. inr this case the operator, who in turn is nat able to charge it completely to tre pas•er- b',:auw;Ir' of restr ict v IATA ruls and:( a str"3 r ct ' i.ntp.tc 1:.t aI r':, .ri, .z nIswer to 11"5 phooum. !.:o t<. lower thetPil cvi I c o - 1' .r'..- : . most advanced airaraft/engine types, tailored for a special mission. The Present effort is c ttnt!'.lt..the fuel burn aspect, beoause tibs has a dlirtt Cffect .- - cost. The raw material cost nowever, is of secondary iXports...... -...... 3% of the final manufactured part price, i. an attempt to -, 14 -owV .o- tential metal temperature and/or creep strength of the hi . ' '" vv] utilizing the latest metallurzical such : dlre.t- Zdevelotmerts single crystal turbine biades, blades witi. a ,i. -1i-.. 0-, metallurgy and composites.

The use of these advanced techniques however, tenes to for" a vicioe s ,io'l,. 'ne the same basic materials are used, even in a more coitiplicatted corpra'iL 'n a'd 'i ficult or not at all repairable; resulting in higher engi:-ne part c~oosts. The lat.itst devti meat, the use of ceramic materials 1 not yet ready for commercial introduction. Howevr, the operator is interested in the capabilities of those materials since another improvement in SFC might be expected because of film cooling deletion and higher turUine inlet t e-!per'trrea mll'wrdrtsgethx- wi~h a -elat4½v cheap basic material. Also for that. coen- buster app) ication (major cobalt consumer) ecramics mrignt `-o a future - ..

MAINTENANCE CONCEPT DEVELOPMENTS

With the commercial introduction of the gasturbine enginae for civii rcroft proiu ti<,: the principles of the military maintenance concept were adopted.Tlis hard time c,,ncke.t started out with the requiremwent to overhaul the complete engine ev, ry p5• iperi'•i :. ur',P Complete engine overhaul was applied up to 1953, and r,ýplaced by . atirag iFourl fot l t l :: ot: Vere , ,ra dLi .coi-1 0 ,-c io:. .. !01.". .' ., . . . . .,

Sectional engine overhaul was appl ied in the 'c-)3-19) t1Iao p-r.iod . d engine module-oriented shopvisit concept. This corc~pt, still basalJ cl r' ti!z.es, ta. dedicated to the sequence of the shopvisit, e.g.: the first slopvisit in: ''pr.ves inspection and repair of the turbine section and visual inspection of the reWIairinc engine sections. The second shopvisit calls for visual inspection of the lo'W prt'sos-re ecm- pressor and overhaul of the crt'ini:lri:s stctio:;. The third spvtcit o..rl'cern-l P the low pressure compressor, inspection and repair of tno turtýe st-ct i,1n and a visual inspection of the remaining engine sections T:sii tt, ''ass repeated uip till the sixtIn shopvisit, whn,:l a complete engine overhaul was applied and the sequens,- rc-nitiated.

With the introduction of wide-body aircraft in our il,_'et in 19(.", t.lo 1il:int-rt*aLce approach was changed to an "on-condition" modular engtine maintenance Coniee ;, based ott engine-hour thresholds for each individual engine module at the cement of sio'vi sit. !I- condition maintenance, which is still in use, is based on. an active line rroznter'.nct- .- nitoring and inspection program aiming at detecting discrepancies inl an early Stageo, los avoiding major secondary d:t'. The following evitr!'t for engine removal apply: - Accute (lost flight crew trig-gers confirmed by a maintenance check) - Def-rrnd (Most maintenance inopection triggers with no :ireet failure ris-k),(- - Other (Convenience reasons, campaig-.ns or FAA limits If any) Upon removal the worksaope is based on individual module thresholds each with an upK•'r and a lower threshold limit with a fixed delta. The escalation of tihe threshold ia based on a number of two samples inspected and overhauled, selected Pr-om the eiiginv iivtrntory exposed for shopvisit between the upper and lower threshold and/or two nsruplt'c with a forced inspection and overhaul requirement whenever the upper thireshold i, ;a;stned. All shopvisits without a sample requirement will be truatea according to workscope otcnpi'led by an Engine Steering Group considering shop Inc, sir inspection results, line mainterlarle reports, operational trend date and modification requirements. Above menrtiontd- Yorocedoreý is developed from past experience gained by the operator and an active design approach of the engine manufacturer in constructing a modular engine suitable for thi maittittlli ,'it concept.

This present maintenance concept does not provide the optimum Utilizatiorn of thet- inldi- vidual engine for it was assumted that the longest on-wing time (low shopvisitratev ) lot, resulted in the lowest overall maintenance cost. Thb>., however, waso il gýenleral trut' f,-r the time period in which no material and fuel .hortage! was evideni. Tw. cetiditions n.uire close control in order to determine the optimum workscope for the eng-ine. - 1. Relation SFC versus shopvisit cost at various chopviaitrates. In general the repair cost tends to go down with a low shopvisit rate-, however, tire lout cost tends to increase under the same conditions. 2. The potential time to wo for each engine is a funl,; ;,'iinof thc, II'fe limitigri parts margin build into the engine. A small margin results in relatively low shopvisit parts costa, for only the limiting parts have to be removed. On the contrary a high shopvisitrate and tie rooturnint7 basic handling costs results. Hlow.ver, a large margin allows a lower sioopviJritrat., but itht • ~~~~~material cost will rise sharply as, a result of' incre-ase'd mat,ýria] io~::t.,... :• utilization of part life. , The optimum of the above mentioned conditions is not ra stanldard ,e' !P;, ti figur" , t function of the "-ngine perfovmanice and material ccondition of each id rivdun1 to. i'..

• • , ,i i I II I i II I 4-3

Future engine maintenance requires a detailed health monitoring and miasion analysis in * order to create a computer model capable of controlling the optimum life of each engine (module), taking into account the actual fuel prices and m.aterial costs. Above mentioned optimalization prQcedures may cause a change in the on-conditiCn maintenance cinoept, 'y acting- on predicted conditions rather than actual failures.

LIMITATIONS IN INSPECTION CRITERIA The condition monitoring of the engine is based on two routines, both aiming at an optimum use of the propulsion system and life 1imitingF parts. - Performance contr1l:. Observation * Acceptan-Lce testing * Trending against a baseline

- Mechanical integrity: Visual checks Boresc ope inspections Oil analysis Non-destructivte testing Radio-isotope insoeectionz *Ultrasonic inspect i-is Condition monitoring of the high temperature parts is limited to detecting exhaust temperature trend shifts at a reference power setting and visual o-servation, mainly b,,, using boresope equipment as an aid. Consider - tne trending as an early warning feature in a more detailed vi uai inspection, it may be conclud-d that the mech.anical integrity ,,f the critical nigh, temperature parts are only as severely i:rspected as the visual capability of tha inspeotcr allows. This is n. problem when ., I ±t- for obvious defects such as missing/bruken or burned/heavy er oded engine parts, hiwever, small cracks and "-minor defects in oritical locations are hard t- detecf . Even for the. poisit routines restrictions are erin 't•tlt..d with respect to determining the remaining tnickness of the thermal coating on the turbine blades and vanes The present pro iedre to strip and re- coat the turbine blade and vane surfaces at :- ivrtni;threshold value gives r.- informaticn with respect to the real coating condition as it is. }'H,ýve'r this re 'cating process may .awv an impact on the remaining fatigue and rupture strength.

0 1 0_From in -p.ratinal point c. view it might be •ncniuded that the less severc y an engine is used, the lower the fatique chance and ieterioration will be. Because the take- off mode of tho aircraft has the largest impact on the life of the engine, derating was Sintr ,duced ac a first attempt to comply with the "--.7eot.ie t, !onrerve on maintenanco ,-cst. Deratinu is presently applied whenever feaL'i'i hlstvv¼r, it is not yet an act vly tool to Oslo:l' ''C'

FUTURE CONTROL OF HIGH TEMPERATURE FARTS Since appropriate methods are not yet available to control the basic engine behaviour in a more sophisticated way, a mathematical method to count the severity of the cycles in a flight mission was developed for military use. Although less severe in respect to the military missions, the routes flown by the civil operator could be subjLcted to the application of those mathematical routines. The first attempt to, utilize this approach on civil aircraft, conducted by SAS and supported by Pratt & Whitney, is based on a mission analysis program using statistical data. The severity factor defirion, based on actual data has not yet been evaluated, but future on-board recording systems like th, AIDS on the AIRHUS Al10 are defined in such a way that all corne.r point conditions are recoroed and thus available for more detailed severity analysis. For the A310 the engine manufacturers made commitments to 'develop an integrated refined cycle counting program as part cf the engine condition monitoring package. The operator needs the manufacturer's know-how concerning the signifioant temperature and loading characteristics of the selected materials and the resulting impact from exposure to oper- "ating conditions in ter-- if low cycle fatigue and thermal cycles. 2-.trol of the mechanical condition of the engine parts, combined with a module oriented performance condition monitoring program such as Gas Path Analysis applied ,w.w:,,[n-aig, ris: a mathematical program providing valuable information in respect to the actuai engine, nodule nd sensor condition, provides an overall :isoet.m:cnt of the engine and 1.' parts condition.

In the operators opinion these new mathematical approaches are a potential feature to control the advanced-design and extremely costly high temperature parts and fuLel burn in order to utilize the optimum life of the engine. It may be concluded that the civil operators are forced to ie-aou ,warc of the nted to create or obtain an effective maintenance tool to further control the behaviour of the engine in order to optimize the material uses and fuel consumption, all. aiming at the objrective to keep air transportation economically justified. see-m C*OO NEIT EI1Eull1 FOR CAUSE

J. & HWti, J., C.>G.0Aa, Jr. M. C, VanWusdeomam.• 0. L S.ir.

Pratt & Whitney Aieuft Geaup Government Prtducts OlvialO P.O0. Box 2M Iwet Plan Beach. Florida u. S. A.

ABSTRACT

Traditionally. cyclic life limited gas turbine enpine components have been retired from service when they reach an analytically determined lower bound life where the first fatigue crack per 1000 parts could be expected. By definition, 99.9% of these components are being retired prematurely as they have (as a population) considerable useful life remaining. Retirement for Cause (RFC) is a procedure which would allow safe utilization of the full life capacity of each individual component. Since gas turbine rotor components are prime candidates and are among the most costly of engine components, adoption of a RFC maintenance philosophy could result in substantial erngine systems life cycle cost savings. Two n ajor technical disciplines must be developed and integrated to realize these cost savings: Fracture Mechanics and Nondestructive Evaluation. This publication discusses the rmethodology and development activity required to integrate these disciplines that provide a viable RI'C syhtem for use on military gas turbine engines. The potential economic benefits of its application to a current engine system are also illustrated.

INTRODUCImmO

Historically, methods used for predicting the life of gas turbine engine rotor components have resulted in a conservative estimation of u.reful life. Most rotor components are limited by low cycle fatigue (LCF), generally expressed in terms of mission equivalency cycles or equivalent engine operation hours. When some predetermined life limit is reached, components are retired from aservice.

Total fatigue life of a component consists of a crPck initiation phase and a crack propagation phase. Engine rotor component initiation life limits are analytically determined using lower bound LCF characteristmcs. This is established by a statistical analysis of data indicating the cyclic iife at which I in 1000 components, such as disks, will have a fatigue induced crack of approximately 0.8 mm length- By definition then, 99.9% of the disks are being retired prematurely. It has been documented that many of the 99 remaining retired disks have considerable useful residual life. Retirement for Cause would allow each component to be used to the full extent of its safe total fatigue life, retirement occurring when a -iuantifiable defect necessitates removal of the component from service. The defect size at which the component is no longer considered safe is determined through nondestructive evaluation and fracture mechanics analyses of the disk material and the disk fracture critical locations, the service cycle, and the overhaul/inspection period. Realization and implementation of a Retirement for Cause Maintenance Methodology will result in system coat savings of two types: direct cost savings resulting from utilization of parts which would be retired and consequently require replacement by new parts; and indirect coot savings resulting from reduction in use of strategic materials, reduction in energy requirements to process new parts, and mitigation of future inflationary pressure on cost of new parts.

RETIREMENT FORt CAUSE METHODOLOGY

Philaeophy - The fatigue process for a typial rotor component such as a disk can be visualized as illustrated in Figure 1. Total fatigue life consists of a crack initiation phase followed by growth and linkup of microcracks. The resulting iacrocrack(s) would then propagate subcriticaLly until the combination of service load (strme) and crack size exceeded the material fracture toughness Catastrophic failure would result if the component had not been retired from service. To preclude such cataclysmic disk (and possibly engine) failures, disks are typically retired at the time where I in 1000 could be expected to have actually initiated a short (0.8 mm) fatigue crack. By definition when 99.9% of the retired disks still have useful life temaining at the time they are removed from service. Under the retirement for cause (IU)C philosophy, each of these disks could he inspected and returned to sevice. The return-to-service (RTS) interval is determined by a fracture mechanics calculation of remaining propagation life from a :rack just small enough to have been missed during inspection. This procedure could be repeated until the disk has incurred measurable damage, at which time it is retired for that reason (cause). Retirement for Cause is a methodology under which an engine component would be retired from service when it bad incurred quantifiable damage, rather than because an analytically determined minimum design life had been reached. Its purpose is not to extend the life of a rotor component, but to utilize safely the full life capacity inherent in that component

This philosophy in itself is not new, and has been used successfully in the past for loi.stresed components which were generally durability, not fracture, critical. For this discussion, fracture-critical components are those components whose failure would likely result in engine power loss preventing sustained flight through single or progressive parts failure. Durabilioy critical components are those parts whose failure would result in a significant maintenance burden but would not likely result in a flight safety problem- The advent of high thrust-to-weight engines with sophisticated (and costly) rotor components made from critical strategic materials, coupkd with advances in the understanding of the fracture processm and improvements in nondestructive insap-tion, has prL.!aptad a renewed interest in applying this philosophy to fracture critical components.

Rsiobkasl LiVe - Foe aimplicity. this scction will consider component life limits which have been determined from the crack initiation characteristics of the specific disk enateril, and will not address 'tw problem of intrinsic crack-life defects. While damage tolerant "conceptsare utilized in some instances to establish life limits, the majority of components in current gas turbine engines have had life limits set by an initiation criterion.

All fatigue data have inherent scatter. The data base used for design life analysis purposes must be applicable to all disks of a given material, and therzfore includes test results from many heats and sources, Data are treated statistically as shown schiematicall, in Figure 2. The distribution of life, defined as the number of cycles necessary to produce a crack approximately OASRmin tong, is obtained for a given set of loading conditions (stress/sLrain, time, temperature)- As can be seen, the t 2w bounds, which contain 96% of the data. may span two orders of magnitude in fatigue initiation life. Critical Fracture

CI MU

Residual Life Initial "Life

Initiation Propagation Total Fatigue Life

FD 220425 j' sgtir' Ta7'I l "utigu" lit,' S.e'gm,'nt&d Inri Stage, f ( ruc, Di'vel.,prno'nt, •ui'C i Gro1trit n th anltd Fintal Froacurt-

LB UB

S(3)+ - 0 0 (3)-

j t iOlll i i " 103 104 101 106 Cycles to Crack Initiation iS4045A Figure 2. Material Data Stcatter Results in Conseruative Life Prediction

When considered with other uncertainties in any design system (e.g., stress analysis error, field mission definition, fahrication deviations, temperature profile uncertainty) the final prediction is made for disk crack initiation life for an oceurrence rate of I in tO(1) disks. It is at thi, life that all !.CF-limited d;-ks are removed from service. This procedure has successfully prevented catastrophic in-service failures However, in retiring 1000 disks because tne may crack, the remaining life of the 999 good disks is not utilized. The amount of usable life remaining can be significant, as shown in Figure 3. Over 80% of the disks have at least 10 lifetimes remaining.

The means of extracting the remaining useful life from each disk must he safe to avoid 'atastrophic failure. This is done by determining the disk crack propagation life (Ni (at every critical location) from a defect bare,v small enough to he missed during inspection. The lI'S interval is then calulated by conducting a lire cycle cost (LC(') analysis to determine the most economical safety factor (SF) to apply to the shortest N, iHITS interval = N,jSFt. Cost vs SF is plotted for each individual ditk and con•bined to determine the most economical interval to return an engine sr engine module for inspection. An example is shown in Figure 4.

Ideally, the first required disk inspection is at or near the end of the analytically determired crack initiation life. Only one disk in 1000 (of that type inspected should have a crack and he retired. The remaining 999 ttould be returned to service for tihe calculated RTS interval. This process is repeated at the end of each HTS interval until a defect is found whose residual life would not allow safe completion of the RTS interval, or conversely, would require such a short RI'I interval that it is not economically or logistically feasible. Then, at the end of each R'PS interval unsafe disks are retired and the others returned to service. Figure 5 illustrates how the residual life is, xtracted from each cisk after the analytically determined crack initiation life is used.

" ' ' ' ' ' ' I i I i i i .... ~ r ...... F •'t .... A S5-1

SENGAE COMPONENT RETIREMENT FOR CAUSE by

J. A. Harris, Jr., C. G. Annia, Jr. M. C. VanWanderham, 0. L. Sims Pratt A Whitney Dircr&nt Group Government Producta Divilion P.0.Obox 291 We.l Palm Beach. Florida 3302 U. S. A.

ABSTRACT

Traditionally, cyclic life limited gas turbine engine components have been retired from service when they reach an analytically determined lower bound life where the first fatigue crack per 1000 parts could be expected. By definition, 99.9% of these components are being retired prematurely as 'hey have (as a population) considerable useful life remaining, Retirement for Cause (RFC) is a procedure which would allow safe utilization of the full life capacity of each individual component. Since gas turbine rotor components are prime candidates and are among the most costly of engine components, adoption of a RFC maintenance philosophy could result in iuhstantial engine systems life cycle cost savings. Two major technical disciplines must be developed and integrated to realize these cost vavings: Fracture Mechanics and Nondestructive Evaluatiun. This publication discusses the methodology and development activity required to integrate these disciplines that provide a viable RFC system for use on military gas turbine engines. The potential economic benefits of its application to a current engine system are also illustrated.

INTRODUCTION

Historically, methods used for predicting the life of gas turbine engine rotor components have resulted in a conservative estimation of useful life. Most rotor components are limited by low cycle fatigue (l.CF), generally expressed in terms of mission equivalency cycles or equivalent engine operation hours. When some predetermined life limit is reached, components are retired from service.

Total fatigue life of a component consists of a crack initiation phase and a crack propagation phase. Engine rotor component initiation life limits are analytically determined using lower bound LCF characteristics. This is established by a statistical analysis of data indicating the cyclic life at which I in 1000 components, such as disks, will have a fatigue induced crack of approximately 0.8 mm length. By definition then, 99.9% of the disks are being retired prematurely. It has been documented that many of the 999 remaining retired disks have considerable useful residual life. Retirement for Cause would allow each component to be used to the full extent of its safe total fatigue life, retirement occurring when a quantifiable defect necessitates removal of the component from service. The defect size at which the component is no longer considered safe is determined through nondestructive evaluation and fracture mechanics analyses of toe disk material and the disk fracture critical locations, the service cycle, and the overhauVinspection period. Realization and implementation of a Retirement for Cause Maintenance Methodology will result in system cost savings of two types: direct cost savings resulting from utilization of parts which would be retired and consequently require replacement by new parts; and indirect cost savings resulting from reduction in use of s'.r..tegic materials. reduction in energy requirements to process new parts, and mitigation of future inflationary pressure on cost of new parts.

RETIREMENT FOR I'AUSE METHODOLOGY

Philosophy - The fatigue process for a typical rotor component such as a disk can be visualiTed as illustrated in Figure 1, Total fatigue life consists of a crack initiation phase followed by growth and linkup of microcracks. The resulting macrocrack(s) would then propagate subcritically until the combination of service load (stress) and crack size exceeded the material fracture toughness, Catastrophic failure would result if the component had not been retired from service, To preclude such cataclysmic disk (and possibly enginel failures, disks are typically retired at the time where I in l00X) could be expected to have actually initiated a short (0.8 mi) fatigue crack, By definition when W9.9% of the retired disks still have useful life remaining at the time they are removed from service. Under the retirement for cause (RFC) philosophy, each of these disks could be inspected and returned to sevice. The return-to-service (RTSI interval is determined by a fracture mechanics calculation of remaining propagation life from a crack just small enough to have been missed during Inspection. This procedure could be repeated until the disk has incurred measurable damage, at which time it is retired for that reason (cause). Retirement for Cause is a methodology under which an engine component would be retired from service when it had incurred quantifiable damage, rather than because an analytically determined minimum design life had been reached. Its purpose is not to extend the life of a rotor component, but to utilize safely the full life capacity Inherent in that component, This philosophy in itself is not new, and has been used successfully in the past for low-streseed components which were generally durability, not fracture, critical. For this discussion, fracture-critical components are those components whos failure would likely result in engine power loss preventing sustained flight through single or progressive parts failure. Durability critical components are those parts whom failure would result in a significant maintenance burden but would not likely result in a flight safety problem. The advent of high thrust.to-weight engines with sophisticated (and costly) rotor components made from critical strategic materials, coupled with advances in the understanding of the fracture processes, and improvements in nondestructive inspection. has prompted a renewed interest in applying this philosophy to fracture critical components.

Redka -ft -- For simplicity., this section will consider component life limits which have been determined from the crack initiation characteristics of the specific disk material, and will not address the problem of intrinsic crack-life defects. While damage tolerant concepts are utilized in some instances to establish life limits, the majority of components in current gas turbine engines have had life limits set by an initiation criterion.

All fatigue data have inherent scatter. The date base used for design lift. analysis purposes must be applicable to all disks of a given material, and therefore includes test results from many heats and sources. Date are treated statistically as shown schematically in Figure 2. The distribjtion of life, defined as the number of cycles necessary to prodluce a crack approximately 0.08 mm long, is obtained for a given set of loading conditions (stress/strain, time, temperature). As can le seen, the ± 2w bounds, which contain 95% of the data. may span two orders of magnitude in fatigue initiation life. 5.3 Gooo

800

00 S600

S400 z 200 oil H1II n r n 10 25 50 75 100 125 150 175 Calculated Lifetimes Remaining FD 1341"7?

FigLure .1. The Majoritv of Disks Have Useful Life After Retirement

High A1or Module: Cost . Optimum V Of SF Range Failure .

Dik

"3 .. Disk 2

Cumulative Cost of 'Tt Frequent Inspections

Increasing Risk Increasing Safety Factor

4. Safety Factor is Determined from an Economic Balance Between SFigure High Cost of Failure vs Cumulative Cost of Frequent Ins.pections 5-4

' Critical X

0 40,

Retire Disk

'feSa aeturn-t0-Service Intervals•

Time

FD 220427 Figure 5. Rasicli retrnient for oaumc cincept

In practice, gas turbine engines are overhauled periodically prior to reaching the crack initiation life limits of their rotor components. This type of procedure could also be used at these overhauls to prevent failure from rc;:we defects. A l(Ft maintenance scheme could he tailored to be compatible with, and an extentsion of, the naintenance "rythm" for a givenl engine system. Initially, it is anticipated that components selected for RFC would he components that already receive inspections at scheduled overhaul periods. These periods are some increment of the expected crack initiation life; therefore, present methods for predicting crack initiation life are adequate. Because of this, improved accuracy in knowing the initiation life and understanding ehe initiation phenomena, while desirable, is not absolutely necessary. Also, it is anticipated tiat upon initial implementation of HI(W, any component found to have a service induced defect of any detectable size would be retired ewvn though a crack proipugation analysis would ind:'at, sufficient ri-siduul lift, for additional RTS intervals, As field service experience builds confidence, or as specific system supportability problems dictate, utilization of the total residual (propagation) life in setting RTl'S intervals could he done. However, assuming that only the initiation lifetimes shown in Figure 3 were utilized, significant economic benefits would be obteined.

TECHNOLOGY DEVELOPMENT REQUIRED

As Figures 4 and 5 illustrate, RTS intervals are based on two broad technologies: nondestructive evaluation (NI)E) and applied fracture mechanics, and evaluated based upon economic factors.

Fracture mechanics must provide an asnessment of the behavior (if a cracked part should it pass NDE with a defect just below an inspection limit. To assure safe return to service of a part which may contain a small crack, an accurate crack propagation prediction is imperative. Recent strides in applied elevated temperature fracture mechanics (References 1, 2, 1, and 4) have provided the necessary mathematical description (models) of basic propagation, i.e., crack growth under conditions of varying loading frequency Wel,stress ratio (ft, and temperature IT). Further work (References 5 and W) has expanded this capability to include loading spectra synergism, i.e., crack growth subjected to (frequent) periodic major load excursions separated by a small number (10-50) of varying subicycles. It is important to note that a typical mission loading spectrum to which gas turbine engines are subjected bears little resemblance to that sxpeslenced by air frames, and therefore different predictive tools are required for each (Reference 7).

Referring again to Figure 5, it is seen that accurate propagation predictions constitute a necessary, but not sufficient, condition for the implementation of retirement for cause RFW). The (,61er requisite technology is high reliability nondestructive evaluation. NI)E must provide the means of screening disks with flaws that could cause component failure within an economically feasible RTS interval. Insufficient NDE reliability has been a major argument against implementation of an RFC maintenance program, NIlE capability with acceptable flaw detection resolution has been available for some time (References 8 end 9), but adequate reliability of flaw detection has been lacking (Reference 10). Complementary inspections and improvements in NDE single inspection reliability thy automation), can provide the required reliability for many gas turbine engine components to economically utilize the RFC maintenance concept.

The fracture mechanics approach to estimating component service life is based on the assumption that materials may contain intrinsic flaws, and that fatigue failure may occur as a result of progressive growth of one. or more of those flaws into a critical crack. Thus, the prediction and monitoring of crack growth as a function of time (or cycles) becomes one of the basic requirements of the analysis system. To utilize such an approach in practice requires quantitative information on component stress, materials characteristics•, and nondestructive evaluation capabilities. Much iif this information cannot ie defined as a single value, but must lie described byi a probability distribution. Two examples are: the probability that a flaw of a given size will exist in virgin material, or the probability of finding a given flaw size with a standard inspection procedure. In order to obtain a deterministic fracture mechanics life prediction (given these distributions), the conventional approach has been to use worst case assumptions for all parameters. PEr pliying all worst case assumptions (deterministic) necessarily results in a conservative estimate for the service life of the component.

To circumvent this difficulty, the problem can be treated probabilixtically. A closed-form solution, whuih takes into account all the required probabilities, is far too complex to he practicable. An alternative solution is to employ romputeýr simulation tec-hniques. A probabilistic life analysis (References 11 and 12) would use a distribution (if flaw sizes. Tibis type if analysis rehul' it-,failure probability as a function of time, Includes NDE reliability, and allows selection of an HI'S interval to obtain an acceptable (1, w) lailure probability with realistic NDF. reliability.

Both deterministic and probabilistic methods could provide some of the NIlE reliability through multiple inslections and/or through higher NDE limits due to shorter RTS intervals. Since mavy NI)E errors are the result of human frailty. Inultiple inspections and automation can enhance detection reliability. A priibabilistic life ana:ysis system, however. would have the ability tit acc•ianm"loiale NDE reliability (probability of detection versus crack length) distributions and assess their effect upon lFt" efficiency. t)bviously, high reliability NDE is desired tWoptimize the economic benefits of RFC. THE RFC IROCWEDURE

The IFC flow chrt (FiSure 6) illusrates a simplified view of how this maintenance concept can be utilized. When an engine (or module) is returned for , iasconomic-sena aalys it a pe•rkormed. (- the engine or module (i,*.. fen, ooupw , bigh 0rban. o low turbine) identified as a participant of the RFC maintenance program. if the module har already been in service for several inspection intervals, the probability of finding cracked parts may be great enough to make reinspection economically undesirable and specific components of that module are retired without being inspected. This is determined by the economic analysis at decision point one and is one of three possible decisions. An unscheduled engine removal (UER) may bring a module out of service that is more economical to return to service for the remainder of its inspection interval than to inspect and recertify it for a new full interval (the second possible decision at point one). The remaining cho~ce at point one is to tear down the module and inspect the parts. During inspect-an, there again are three possibilities (decision point two). If no crack is found, the part is returned to service. If the disk is found to be unsafe, it is retired. The third choice is to investigate modification or repair of a flawed pert. An economically repairable part may be repaired and returned to inspection (decision point three).

Initial Engine/Module Proremtj -I

-e ~Fiod Operatio

,,m-•~~pe cnomc=opfi

Shop Visit Overhaul

Anan etRtInspection

Insec RepairableI

Unsafe Retire

FO tenA

Figure 6. Retirement for Cause Flow Chart 5-6

AN APPU.CAhIO OF FhTUhlN FOR CAUSE

The application of a RFC maintenance approach to a military gas turbine engine has been studied (Reference 13l, and development activity is undereay to reduce the RFC concept to practit. The dnmo-ustratwo and first Umplmeatatoa system is the United Stats Air Force (USAF) FIO0 engine. This engine is a"1 augmented turbofan engine in the I 1OkN thrust cls with a thasis-to-weight itsie in excess of 8 to 1. The engine is currently in operational service around the world in the twin-engine McDonnell-Douglas F-15 and single engine General Dynamics F-16 fighter aircraft.

The FI00 is an axial flow, low-bypass, high compression ratio, twin spool engine with an annular combustor and common flow augmentor. ft has a three-stage fan driven by a two-stage, low-pressure turbine and a ten-stage compressor driven by a two-stage, high-pressure turbine, The engine consists of five major modules: fan, core (compressor, combustor and compressor-drive turbine), fan drive turbine, augmentor and exhaust nozzle, and gearbox. Each module is completely interchangeable from -!ngine-to-cngine at the intermediate maintenance level. The modular approach was selected so that either functionally or physically related parts can be removed as units.

The objective of the study program was to determine the feasibility of applying a RFC maintenance approach to the USAF Fi00 engine. The study was directed primarily toward rotating components of that engine, specifically the various fan, compressor and turbine disks and the spacers/air seals that comprise the prime rotor structure. The effort addressed the following five major areas:

"* Definition of an RFC methodology "* Evaluation of the disks and other appropriate engine rotor components for RFC applicability "* Assessment of the nondestructive evaluation requirements for implementation "* Establishment of a component ranking for development priorities "* Fstablishment of development plans leading to implementation.

The methodology has been discussed in the preceding section of this paper. The life analysis of every major component was reviewed using deterministic lifeing procedures. Based upon this life analysis, individual component coat, and a 15-year average engine system life, 21-candidate r-itor comnponents were selected. The components include: the three fan disks, four of the compressor disks, the four turbine disk%, and ten airseels/spacers from the fan, comprecsor and turbine. These components are all produced from wrought titanium and nickel base alloys.

Each critical area of each component was then analyzed for residual life to establish the type of defect and the detection requirements range for optimal emonomic benefit. A composite sketch of typical rotor components is shown in Figure 7. As can be seen, the configuration of these components is complex, and innovative techniques are required to enable reliable inspection, The types of defects to be detected include: corner, surface, through-thickness, and internal flaws. Techniques currently exist, such as eddy current, ultra sound, penetrants or proof tests, with the ability to detect these flaws in the sizes anticipated, The major problem to be solved is detecting these flaws in a high-volume maintenance environment with sufficient reliability to enable RFC to be economically viable.

The ranking of components, or groups of components, was based upon three factors: life cycle cost impact, NIDE requirements, and an assessment of where the state of the art in applied fracture mechanics is at this time. The final ranking of components for the FI00 was done by engine module because NIlE requirements are similar among components of the same module; it is economically and logistically impractical to return units of less than a complete module to an engine overhaul center. The development priorities are: compressor components, high-pressure turbine components, low-pressure turbine components and fan components. This ordcr was chosen based upon realistically meeting the NDE goals and the life cycli, cost savings per module. In all probability, the chronological order in which RPC would be applied to this engine is: fan, low-pressure turbine, compressor, and high-pressure turbine based upon present technology.

From the results of the first four activities, a preliminary plan was developed to identify the technology and other activities required to enable implemcntation and operation of RFC at an engine overhaul center, and the time pharing necessary for a target application date of 1985. The logic upon which the plan was developed is shown in simplified form in Figure 8. There are five sequential steps inherent in the planning process: (1) development of the required technological and management tools, (2) establishment of inputs to and exercising the tools, (3) demonstrating the tools, (4) evaluation and documentation of the tools, and (5) implementation end use of the tooIs. The resulting development plan is shown schematically in Figure 9 and addresses the technology developments discussed in previous sections of this paper. At the present time, these activities are underway and proceeding according to the planned schedule.

Benefits of Retirement for Cause - The assessment of the benefits of a Retirement for Cause maintenance approach to a gas turbine engine is contingent upon many assumptions. Thes assumptions include: fleet size, anticipated usage rates, usage life, inspection interval, labor cotts, parts cost and many others,

To quantify the benefits of this maintenance concept for the USAP P100 engine, life cycle coat analyses were conducted. These analyses determined the change in life cycle costs of the FI00 engine that could accrue based upon implementation of an RFC maintenance procedure in 1985 as opposed to a continuation of current or baseline maintenance practices.

The life cycle cost benefits amount to an approximate U.S. (1979) $250 million savings over a 15-year period. In comparison to the investment required, the development and implementation of RFC are extremely attractive. Knife-Edge Seals

Radial Cooling Hole/Balance Cut Blade

Blade Retaining

Blade Slot Seal/ Rim Radial BaeSpacer Cooling Ai, Hole

Rim -Web eb Cooling • -Oil Aii Hole

Balance Hole Fl ng Spacer -/• W eb

WebW S• Arm/Flange)

Dissk (Web/ Bre

gCooling

Air Hole Engine Ceoter Line Air Hole•

Figure 7. Composite Sketch of Typical FlOO Rotor Components and Fl4w Types (Not All Features on All Parts)

Probabilistic Fracture Mechanics NonDestructive Evaluation "*Technology Base - NDE Characterization ""Component Life Analysis o System Integration "•RFC Plan - System Control I/ Life Assessment Systems : RFC/NDE Systems

RFC Procedures

Methodology Demonstration

Implementation

FD 520428

Figure 8. Technology jDeelopment Required for Engine Component Retirement for Cause 1. Conet D iton

ilU. LI. Aaaamlint Systems A. Lira Predic•tion RrAarch .. P

8. Methodology Deve mIIe.nt C. Component Life Analysis D. RFC Plans =

II. NDE Systems A. Quantitative NOE Research

B. NDE System Design

C. NDE System Development

IV. Methodology Demonstration A. Component Veiflication B. Engine Testing =

V. Implementation

Vi. Documentation and Coordination

VII. Sustaining Engineering Activities

1979 1980 1981 1982 1983 1984 1985

Figure 9. Dervelopment Plan for Engine Cunpocricnt Retirement for ('ause

CONCLUSIONS

Realization and implementation of a Retirement for Cause Maintenance Methodology will result in system cost savings of two types: direct cost savings resulting from utilization of parts which would be retired and consequently require replacement by new parts; and indirect cost savings resulting from reduction in use of strategic materials, reduction in energy requirements to process new parsl, and mitigation of future inflationary pressure on coat of new parts. With the worldwide concern over the availability of strategic materials, the resultant reduction in requirements for, and conservation of critical resources may eventually become as iarge a factor in applying RFC to specific systems as the direct cost savings upon which the decisions are currently based.

'The methodology and procedures described herein are applicable tA)systems other then the FO00 engine. A curs)ry review of other gas turbine engines indicates that the RFC maintenance cor.cept is generic and has direct applicability to rotor components of those engines. In fact. the methodology has broad applicability to other types of engine components, and indeed, to systems other than aircraft gas turbine engines. The decision to apply RFC to other components or systems would be based upon economic factors, predicated upon the remaining anticipated service life of that system, and it could be a viable maintenance concept for life limited components of all types.

REFERENCES

1. Annis. C. G.. Jr.. et a&.,Pratt & Whitney Aircraft. "An Interpolative Model for Elevated Temperature Fatigue Crack Propagation," 1976, Air Force Materials Laboratories TR76-176. Part I.

2. Wallace, R. M, et a&.,Pratt & Whitney Aircraft. "Application of Fracture Mechanics at Elevated Temperatures," 1976. Air Force Materials Laboratory TR76-176, Part I1.

:3. Sims, D. L.. et al.. Pratt & Whitney Aircraft. "'Cumulative Damage Fracture Mechanics at Elevated Temperatures." 1976, Air Force Materials Laboratories TR76-176, Part Ill.

4. Sims. 1). L., Pratt & Whitney Aircraft. "Evaluation of Crack Growth in Advanced PiM Alloys," 1979, Air Force Materials Laboratory TR79-4160.

5. larsen. 3. M., et al-, "Observation of ('rack Retardation Resulting from load Sequencing, Characteristics of Military (Ga5sTurbine Operation." AS'rM STP.i 14. "Effects of Load Spectrum Variables on Fatigue track initiation and Propagation.- 19&). pp 91-107.

6. Larsen, ,I. M., et al.. Pratt & Whitney Aircraft, "(Cumulative Damage Fracture Mechanics tinder Engine Spectra." 1979, Air Force Materials I.ahoratiries TR79-4159.

7. Annis, C. G. Jr.. "An Engineering Approach to 'umulative l)anmag Fracture Mechanics in G.as Turbine D)isk.," presented at ASME (ass Turbine ('onference. San l Cieg,.CA, March 1979 8. Hyzak, J. M., et al., USAF, "Development of Quantitative NDI for Retirement for Cause," 1979, Air Feoce Materiab Labomrmes TR78-W8'

9, Cargill, J. S., et il., Prstt & Whitney Aircraft, "Disk Residual Life Studisa," I97/• Air Force Materials Lsborateies TR79-l23.

10. Lewis. W. H., et el., The Lockheed-Georgia Co., "Reliability of Nondestructive Inspections.' 1978, Sai, Antonio Air Logistic Center MME 79-6-38-1. 11. Annis, C. G.. Jr., et al., Pratt & Whitney Aircraft, "Probabilistic Fracture Mt.- .nics and Retirement for Cause," internal "nnmmunication to be submitted for publication.

12. tau. C. A., Jr., et al., Failure Analysis Associates. "rhe Impact of Inspection and Analysis Uncertainty on Reliability Prediction and Life Extension Strategy," presented at DARPA/AFML Review of Progress in Quantative NOE, La Jolla. CA, July 1978.

13. Harris, J. A., Jr., et al,, Pratt & Whitney Aircraft, "Co'iept Definition: Retirement for Cause of F100 Rotor Components," 1980, Air Force Wright Aeronautical Laboratories TR8O.4118.

ACKNOWLJEDGEMEINTS

The concept of retirement for cause (RFC) as presented in this paper has evolIved based upon the work of many individuals within the industrial, academic and government communities, The authors acknowledge these contributions which have enabled RFC to move towards reality as a maintenance procedure for military gas turbine engines. Special acknowledgement is accorded to Dr. W. H. Reimann. Air Force Wright Aeronautical Laboratories, Materials Laboratory. Wright-Patterson Air Force Base, Ohio, for his technical contributions. management sponsorship and perseverance in guiding RFC of gas turbine engine rotor components from idea to reality, and to the Defense Advanced Research Projects Agency, Washington, DC for their sponsorship of the contract program activities.

io4 6-1

DEFECTS AND THEIR EFFECT ON THE BEHAVIOUR OF GAS

TIURBNE DISCS

Robert H Jeal Head of Materials Engineering Rolls-Royce Limited P 0 Box 31 Derby DE2 8BJ England

Summary

All materials contain defects but recent moves to higher stress levels have lead to an increasing number of these defects being malignant. Unless the method used to life fatigue critical components like gas turbine discs allows assessment of defect presence and behaviour, the risk o' serious failure is dramatically increased.

A method is discussed where defect behaviour can be assessed as part of a total life approach to disc behaviour prediction and is explained together with the effects of differing defect types. Such an approach gives realistic manufacturing standards and controls and leads directly to an 'on condition life' approach.

1. Introduction

Since the realisation, some 30 - 40 years ago, that gas turbine discs were the most critical components in the engine because pieces cannot be contained if failure occurs, various criteria have been used to define both the design parameters and lives of such components. Until the end of the 1950's the main factors considered were creep, for the rims of turbine discs, and Lensile strength, to give a satisfactory overspeed margin without burst, for both compressor and turbine discs.

The lessons were learned early - as can be seen from the failed turbine disc In the Whittle Mk3 engine in the British Science Museum (Figure 1).

In the period up to the mid 1960's failures were due, in the main, to lack of quality control, where the material didn't meet specification, to design problems mainly at stress concentrations (dealt with by geometry modification) or to problems such as corrosion and fretting w:hich were alluviated by material change or coatings. In Britishturbines in particular, little attention had to be paid to low cycle fatigue problems because of the extensive use of martensitic steels where the limitations for o%'erspee$ meant a fatigue life well beyond that practical in service.

This situation changed with the introduction, first of nickel base and then titanium alloys in both turbine and compressor discs where, although creep and tensile criteria still needed observing, it was found that the most important factor governing component design and performance was low cycle fatigue.

The literature published on the subject of fatigue fills many library shelves, but, as applied to gas turbine discs, this approach assumes that materials initiate failure origins that turn into cracks in a way that can be predicted from relatively simple laboratory tests, with a variation in performance that can be handled statistically. This method of life prediction, which Is still used in many areas of the industry, is based upon assumptions that the bounds of the material behaviour can be circumscribed by manufacturing control and the presence of defects eliminated by inspection (Figure 2). This paper sets out to show that recent developments both in materials and design now require a more complex approach to disc performance and associated manufacturing criteria. 2. Basic Lifing Approach

The 1icpDning authorities for civil engines demand that the manufacturer establishes the finite life of some 'best' specimen and then allows 'in flight' use of a proportion of this life to cover the 'worst' expected example of the disc family. A similar lifing system is used for military discs. The exact way of defining the finite life varies from authority to authority (and the mode of establishing that life from manufacturer to manufacturer).

The British authority (CAA) and RollF-Royce have always used cyclic spin pit testing under near engine conditions as the basis of disc hl'ing. Whilst laboratory materials fatigue data is used in the design context it has not nii.-d calculate in-service performance. Such an approach ts based on eareul control of total manufacture Lo make sure that the components tested are in the same family as those in service, together with careful control of 1nfCpeCiron. The end point in life has, in the main, been taken as % burst liýfe rather than 'first crack' . The latter is. rather a nebulous- conc~pt when regarded as a life criterien (Figure 3).

in reality this has lead to a lifing approach which ha,'-depended at, :1 '•r-o •tages of life - ' initiation'f pronpagation' and 'final failure' . At fi"-. th's was, of c.urse, empirical but during the past- 10 years the mater alas -nginer'i, ha'• hn• toc,0hn Iques and under-st andi ng to''o' rol the -:arlouis life ph'±'

3.1 Crack nurrclat on

3. 3A StressDisn'a.'d r'erow'h 3.4 F4inal 1'~ur

:3.1 and 3.2 ar,"e' ''lly von-zd,'red tce'her under lit, 'Iin 'at in' oil'ept. The "en-d pot;0 S ' ross "on"sdel-ac "v rP'a"'r' dr':i"k''e r'"''"an ,oýur-Int . 'ter stage 3.2 ,r .3upondefntIn f"" and",f 'r'-.mater'il ai

3.1 Crack nuchsloiao-'n '.,h - ,f fa'c. ' ' , ,, o !.: ttr'ai una il a phy aa' , 'lnu y or lra k' ,'-.rur;.T hP"'T - a numbe'r o.f dlfferent- wayn:. b'' h at the :.urfai e and within the -. pe; ,a'no'whenhey no,ýn-cchcrr': ct,--c'"are pres4en' or not.

Asý ateri ,1 of higher w &trer h are us~ed they' tend ý'r'-in lo"",l i"-a'ion vpecs L-f behaviou' in this4 stcgr wh4' h 'ncrcan,' the propt-,e'- ''y for ' ir- •rtae fai lure ,(Figtre, . The :I.i'f tE= lia- depend'' upon'"'',-i'ndlng - -: r in i? isiu precc-ice or not, of defec-t) ,-cmeratur' .l,-'-'71''-!0 0'.

-.3.2.. ho c rack he'n .l.i lost''"'' " r' 1 ci's 't : h , ow h oxcrn-d i's ,he "ame- fSai'ors as 3.1 T'hat it local ,-Ci'" f ' I':t' -dra at the- ,'rask which are those of t'e '_n di\'idual grain th.t ra, kke. The 1'f, ýr tlhis ;,has-" c-an yar,' from nil for- ome maerialas ahere lit nucletion pro Is e that

produces grain size '.'raS:,. 'anytr :itc '' ,f 'he mul'1e0-' "' ha'-" whore erolwth ,-.at 'his sje ag involves a -'eruc-' of nucleati''r,'- - each at higlier local atre.: urril the crqok propagntion hrieshAold is reached.

"3.3 On~e the crick can be considered large i.e. above gr-a'n size. its behaviour is dcm-nated by the imp •ozrd ,tress ax'ston in the "',ode niOrmally predictted by the crst-k prepaoigation phasae of fra.-tur-e mochainics analy sis. Out.ide terperature and stress cu 'h behaviour is indepodent of mos-,t paranetere ex -ept basic alloy systes - too".it martensitic steels behave the same, as do nickel and titanium alloys but each gr-oup has significantly different crack propagation rates from the others.

3. .I Thi s sr-ak grows until come final failure !riterion :s ,exeeeded. This, normallsv is dictated by fract',uro tnghnreos criteria, the eutset of unstable crack growth, but can) also be that the section stre:ss exceeds the tensi'le level or that -ome sort of plastic tearing mode occurs. A variety of failure prediction concepts are availault- but final size at burst has little effect on overall life - although it ,b i ,-aucalv has-, insqectoni irmplications.

SThe percentage of ;•fe spent in each of thetse phases varies according to material, stress/atrain level and presence of defects.

4. Dfects

-PDfort: can te c ar-si fied into two tIpec, -

4.1 - Physical dsiioentinuities a.? - ,icro s.tructural deviations that pre10n,1'0 scatrter' itr hohaviour outs'ide that "de:=i gned for.

!1.1 Phy:sical di -.eoont inu i 1 Pes

"This Cat'egar, cover:- cracksi i-c.'nctens, areasý of --ontariinatioi, slag anrd 'thier- tien-coheren t areas '1ihin t he component 0-3

They Sfect component life by either eliminating or severely reducing the 'initiation' or crack nucleation phase. The effects are geometrical;that is, such defects create a disturbance in The section stress field which accelerates the build up of fatigue miamage, leading to early crack formation.

If their presence and geometry is known then their effect can be accounted for in the life method, either empirically by disc testing or theoreticallv by stres calculation and laboratory testing.

4.2 Structure deviations

This category covers coherent defects where any volume within the main body of the material behaves differently under the imposed stress conditions. The defined scatter of material properties depend upon defined bands of microstructure and chemistry.

In volumes of material where the latter factors lie outside the defined limits the material has a different stress-strain-time behaviour which imposes higher than predicted stresses in or about the defeotive area leading to more rapid failure.

Both types of defect are, and always have been, pre-,ent in even the best ceommerciallv m.anufactured components. Thetr presence can, however, either be benign or malignant. depending upon their size and the imposed stress conditions. The traditional 'no defect' approach has relied on eliminating all malignant defects. Current materials and operating conditions are making such an approach unrealistic as defects below NDI method detection limits are in so-e cases nsol malignan:.

F_=. iethods of accounting for defect behaviour

Disc lifing requires an understanding of component life in the terms described in. "section 3 above so that appropriate data and methoýds can be applied to each stage. The effect of various types of defect can then easily be assessed provided t*,er presence is recognised and cisirbehaviour known - at least within bounds, if not "exactly. This approach can then be aplied to develop a first life for an individual component using aodifica%"ons of existing, fatigue life and fracture mechanicsmethods, and to subsequently extend component life in further increments using scne form of inspection - a life on condition or ret irement for cause system.

This overall ascEssment of life also allows the practical considerations of such factors as frettage, corrosion and mechanical damage.

The type of data and approach required for each life stage is discussed below.

5.1 Crack nucleation

Knowledge of the nucleation mechanism is ýli important here - in particular is thi- likely to be associated with a non-coherent defect or does the crack nucleate at some coherent structure origin?

in the former case life declaration for this phase depends, upon definition of defect size and knowledge of the condition.s when the defects become malignant. if they are benign e.g. silicides in Titanium alloysthen their presence ,an, of course, bk igoed xt rapsiat ion. ,f prrovt~ expevrienceý:,t hiehert *:, r,-: levels can be very mis,leading in such cases.7

Definition of defect siize depend-- upon practical knowledge and experionce of various %iN metheds and their interpretation to reality. Figure I Kshow:s the relationship of true size and ultrasonic response for a certain component in Waspaloy. In particular it shows the statistical nature of this type of information - which also applies to eddy current, penetrant or binocular inspectionl of sLurfacce cracks.

Knowledge of defect type depends upon detailed understanding of the total manufacturing sequence so that the chance of getting a certain type If defect in a particular volume of the disc can also be taken into account - another statisticol factor.

Understanding of the defect behaviour usually requires* firt hand ,bservat iOns f behaviour of components containing defects at represexitative stress,!:inc/tempera:ur,, conditions. hometLmtao t',ie shows that the geose! rical effect can be as1osoed by ---3D finite element techni qu (Figure 0i) but more oft en the ,nset of defect Mallnancy and "ts extent can only be determined cap cii,'.tl. Even where many defects are present - as in powder nickel base di:ýcs- - their behaviour can only be statistically assessed by interpolation to engine conditions from laboratory and component tests.

-- Mild ------

Where mnal ;inant do'fct t re not. pret-nt and crack nue atiaIn t0000s at tru•cturaI u ceoIt.: port at' :- to uun-i tl -,uid ,rnd 0:- '- 4 t:.u'.ta..t so that the variation of nucleation sit-s pr'eatfltL "r ,cent rol :w 50.1 ttr Land o-f hehavicur t- fh, aFsumed in the life cal- lam:onr.

Again the approach requc:re: a dt,' ailed knowltedge of the mechanism, snd in: rip-: atI of data. Ext rapolat4,in of experience to higher stress levels can be misleadiapt A number ," oases- have been rv,-orded wh,,re a modes; chanýe in truc Lure has ,r--O: -d an entirely new mode of behaviour (Figure 7).

5.? Small Crack Growth

Whilst this phase, if it exists at all, must be considered separately from 5.1 moot- of the above remarks on defect dependence apply. The most important. factor is nucleated crack size and whether, under the locally applied operating conditions, i.t is capable of propagation or requires a number of renucleation steps - each time from a bigger 'defect'.

As the crack is small its behaviour is dominated by local conditions within the material and so assessment of length of this phase must be statistical.

5.3 Large Crack Growth This phase of life is probably the easiest to assess as the fracture mechanics techniques required are now well known and the scatter of behaviour small. The main factor of uncertainty is in definition of the defect size at the start of the phase - particularly if a sub-surface site is involved.

5.4 Final Fracture

By the time the crack has reached the size for f'nal fracture, growth is so rapid that total life is relatively insensitive to this factor. It is important, however, in assessing the risk of in-service failure which is higher if the final fracture size is similar to the limit of inspection method available.

Special attention has to be applied where change in temperature promotes a step in toughness for steels or causes other such dimsontinuities in behaviour.

The statistical nature of all these phases must be noted and the life declaration for thc component can olnly be expressed in a statistical form -a risk analysis for any given calculated life should bo declared (Figure 8). This involves defining some definite life end point - like ij burst life rather than an ill-defined 'firs: crack critcrkn' or a quasi-quantitative crack size e.g. .75rmm which gives a dramatically different end point depending upon the stress conditions and material involved - in addition to •he problem of coping with 1mm defects using that approach!

Whilst the above has concentrated on material behaviour factors it also has to be emphasised that the approach requires excellent 3D stressing of the component including detailed knowledge of thermal and residual stresses. .Hthout this knowledge of stress levels and gradients - which in itself depenis upon detailed stress strain knowledge of the material - the effect of all important component geometry cat nullify other considerations. 6. Case studies of 'defect' behaviour

The application of this approach to a number of real cases of defect pn,-:ense aLnd behavlour is discussed briefly below.

6.1 Failure of a large Titanium Fan Disc

Conventional lifing of a fan disc had defined a life of more than 10,000 flights with final fracture occurring at the disc bore. Whilst spin pit testing confirmed a high life, final failure occurred from the back face. As investigations were proceeding to examine this discrepancy two in-service failures occurred (Figure 7) in less than 1,000 flights. The lifing method and predictions were based on laboratory specimen testing and spin pit testing of typical samples of the appropriate alloy at 15 to 20 stress cycles/minute. Investigations of the problem (which was difficult as the in- service failures were not recovered in either case) showed that the effect of three unconssidered and unknown parameters had combined to cause early failure. These were :-

a) The manufacturing route of the disc lead to high residual tensile stresses at the b--k fn,- which moved the most critically strezsed position away from thai p red icoted. b) Ingot porosity occurring in the ingot hot-top region wa.t not healed by subsequent metal working and was too fine to be detected by ava•lablei'IDI techniques

c) The material microstructure in the back face region was a corser, lrw -n•- version of that of the typical alloy.which promoted a step change in 'heK behaviour. The material at the failure site showed alignment over -Igntf-car- areas which allowed the small pores to rapidly become large propagating cra*ks and these cracks 7rc-,pc-ato 81 t , . rdci-r:" -F mug - "r ov 'dwell' conditions than that measured under normal test conditIors. 'he neH effect is shown in Figure 9. Application of the type of lifiig ap, r a.-h hn above allowed the problem to be identified and reproduced even though the failures weren't recoverfd,and -solutions to be found. These showed that a safe, albeit short,life could be toleated in the immediate period whils- a lens term solution was reached. It also showed that a material change was required and why.

6.2 Behaviour of a Waspaloy Turbine Disc

A rig test disc. being used fc.r normal life develotment chewed extensve esariv cracking from a sub-surface defect well :nside the ultra --oht acceptance i-t-.'c- (1.25mm fiat bottom hole equivalent).

The defect behaved predictably once it was assumed ,, a cract: (Figure !t Subsequeni investigation established ~hat it_ had g';n al o;1trasont`- response (around .7mm fbh) even though it was large and -alig:-an:. (FIgure '-. To prevent in-service failures a new ultrasonlc !tandard waoo adooted related to real defect size and the component life relat-.d -(. 'he largest undetected defect size - the life i.,method detailed above was aipl e1 .

6.3 Use of powder Ni alloys as engine discs

Consolidated Ni-base powder discs inherently contain defects from cha marufam:urine process. These defects are below the detectability of current -I'D meth-.ds and Ye, they are the fn--or soy.,•orn•i'; compone b..hot- .r (F--re "

A comparison of lifing methods, and acceptable param.eters to:sno.wnr in F~gure ! hr the 'defect'governed behaviour is presented in a Fatqilue curve fort. for compa it_-r withatraditional fatigue a}proc:,h. Application cf the total life sco-.r that the use of such a product can be safe provid ed defect behaviro-ur criterisa at usedj that the extrapolation of traditional fatigue and tensile design criteria -e higher stresses can be very misleading, causing early component replacement at best - in-service failure at worst; and that material development needs to accoon-" for defect tolerance when the alloy/structure compromise is reached.

7. lmplications of Total Life system (fi__ry 14)

Rolls-Royce have been applying this approach of disc lifing in an incea- - comprehensive way over the past eight or nine years - based on extensive 1- n.o -emerials trsting, component spin pit testing and manufacturin.g defect ivt-stigocion. Experience :ias revealed the following advantages:-

a) Identification of quantitative defect standards for all features (includin• welds) which allow realistic definition of manufacturing allowable against component operating requirements

b) 'Hazard review' of manufacturing processes to identify critical phases allcwinc a safe but economic approach to product specifications on the basis of operating requi rements.

c) Safer use of material at high operating stress levels with a realistic approach to design allowables.

It is already allowing a 'life on condition' approach to individual component life extension in service beyond the initial 'family life' as well as realistic use? and definition of inherently defect-imited materials and processes such as powder nickel, castings and welds. Whilst the spread of life on condition applicatuons ar( being held back by NDI capability the total life method still allows a total examination of component manufacture and operation to achleve a safe Pcenonom!,ilife - the only way that the high streng•th marrri ls required by the new efficient aero dynamic designs can be used. th uh( r-krwece rhf he~lp r4i hiV friend- a& claus- stezal ,1.,- rA fr the myctc.ci -. ,f 2,-'' Cmln , 'lcrmont and .,,r D lbhnns for plram; i,-a] lifing an3ssssments.

Jr Vf ,fzor W qoeppner also provided inspiration through the 'ifficult tires. 6.7

Q Ni 0 4 N,

F4,

m m •i

m i

.= -- u-ýý tk!A W'6=It" L

...... ow GuolJ.

U... .

m- Pciis-r- ý=c-mIur=P G.. r ..... - .. n W- B" -rrL_!&-W

-CI tUkAI j T 1 4A C t mC Mr.C"m'> ýL OV<

A.L.Ex-4A=ý C&AI 6-8

CHADlTN.

c c-r')DAT"r E

MRIJ•co.raO an .-- s

.AL- a cm ann~u..,..-r&I ýIL f

Raa.,° - riao-~ . -- ~ oc c.c-. e~iar ,. 04,.U..C

• Na

' C tt 5io-sa mi

:: a)_- I o n O ar ss L o,.p"--o S m.-itt•J - ca.eu-amooiM..s

AL LO'aSAc L-l LwlEc-S.C 6-9

Fig.1 -Failed 'Turbine Disc -Whittle Mk .3 Engine (Courtesy Proc. Inst. Mech. Eng. 1945 152 419) i Engineering premise

All materials are homogeneous, elastic, isotropic media, free from defects and amenable to conventional fatigue cumulative damage analysis within normal scatter limits

Fig.2 The 'Engineers Premise' 1 t,... (-I0

Conventional disc lifing system Data bank -,Stress I Design_I' Spin pUit test Assumed defect I for initial release free after NDT Factor I Ex-service inspect/ spin test I Life increase

Data Bank

Conventional properties SMI 1315S6

Fig.3 - Conventional tifing Syitemn

"- Failure modes Applied stress Applied stress

Discontinuity Discontinuity

Ductile mode Quasi cleavage P mode

IML 11159

kig.4rack - Tip1 .'FalireI inMde-, Forged Waspaloy -Ultrasonic indications conversion factors Large turbine disc

5- Radial (95%)

4ZRadial (50%)

3- S~0

1 Axial (95%) -- -Axial (50%) !jS' - I I 1 5 20 50 90 99 99.9 SML 1575S

Fig.5 - Ultrasonic Response of Defects

, Cr . ~.... a,•r•r .

S d/2 d/2•

0 2S 0 5

Oisoonte - ciu-,m/mIuMe-.m - lhek". --

Fi,.6 - Defect St ress F'fects 6-12

- ok i.

Fig.7 -Effects of a Fan Disc Failure

SRisk analysis

9 Probability of the component containing a defect * Probability of the defect being detected e Probability of the defect being malignant * Probability of the defect being in a particular position and orientation in the component * Relationship of the NDEindication to the real defect size - cut-ups of components containing defects e Probability of a defect in a given position being large enough to cause failure in a given number of cycles or flights a Variability of material properties * Probability of a failure being hazardous * Definition of an acceptable probability of failure (or hazardous failure)

ig.8 - R isk Ana Iys is 6-13

Behaviour of cracks in a fan disc in IMI 685

Initiation phase iPropagation phase

in service behaviour expected behaviour

N SML I1*16

Fig.9 - Crack Progreý,si~on in a Fan Disc

~jCrack propagation from a subsurface defect

Nickel base superalloy disc rig test -073, i -defect size, shape and distribution

IWCLUSION

AftA I FRONT FACE .0611172

Fig.1O Propagation of a Sub-Surface Defect 6-14

SCrack propagation from a subsurface defect nickel base superallay disc rig test

" "" " &' ' During manufacture, a defect ". V,1 was detected in the bore using kil- ,,ultrasonics, at -11dB to -12dB -4 \ ' level "N-*''' ' •- The disc was rig tested until * '~ i'••" failure occurred in the rim ~. region. The bore crack shown •• ". "was, then opened. Ratio of predicted to actual life, using CT specimen data, is 0.86. Ratio of predicted to actual life, using corner crack data, is 1.07 Defect Predicted final Ratio of predicted to actual crack shape life, from striation count is 1.44 ± G.48.

"Fig.1i - Sub-Surface De-fect in Waspaloy

POWDER ASTROLOY Low Cycle Fatigue Failure from Subsurface Defect

1Imm 50 pn

Fig. 12 - Defect Fat igu e rri. in i A.t ir.x- S~ 6 -IS

I POWDER ASTROLOY Low Cycle Fatigue Behaviour at 600 0 C

F RACTURE MECHANICS LIFE 1200 150 pm POWDER DEFECTS

STRESS 1100 DEFECT FREE FATIGUE LIFE RANGE MPa 1000 FRACTURE MECHANICS LIFE 400pm " FATIGUE LIFE WITH 900 ULTRASONIC DEFECTS 150Mm DEFECTS

$00

10 2 10 3 10 4 I0 5 CYCLIC LIFE

Fig.13 - Cyclic Behaviour of a Powder Superalloy

___ Total disc lifing system Data bank Finite element stress/thermal analysis Design- Process/NDT capability Sii=Confirm co egnn ii 0014'n pt est Fracture mechanics |iiii. fconditions",LifeU e • assessment I Factor for initial release Assumes presence I Ex-service inspect/spin test of defects to NDT/ I process capability, Life increase surface & sub-surface I Retirement for cause Data bank Properties Behaviour Service Stress method NDT capability -4 NDT Engine environment Developments 11 ill i M• 11l3$

FIg.14 - lotal Disc ILif'ing System !Ri-

R -1

"RECORDER'SREPORT - SESSION I

Dr M.G.Cockroft National Gas Turbine Establishment Pyestock. Farnborouglh. Hants GU14 6TD UK

Wg Cdr Hedgecock's paper emphasised the value of recording actual engine useage by means of simple low cycle fatigue counters or more comprehensive monitoring systems. The discussion largely concerned points of detail about such systems.

Replies to various questions can be summarised as follows. The LCF counter in its simplest form takes four shaft speeds as input, possibly two speeds from each of two engines. Speeds are converted to stresses and stress excursions during operation are summed to give the equivalent number of zero/maximum stress cycles which dictate disc life. There is no analogy with the "reference stress" concept used in creep life predictions. Many more inputs (e.g. temperatures, are required for the more elaborate systems which would take into account creep life. thf nnal fatigue lift. etc.. and a "hot end monitor" is in process of being developed,

Professor Hoeppner raised the question of the criterion for disc lifmg: whether it is "life to first crack" and whether surface and internal defects are treated similarly. It was pointed out that sophisticated Retirement for Cause inspection procedures are not used for RAF engine discs after they have entered service; lives are set by the manufacturers. Formerly lives were prescribed on the basis of "life to first crack" but the current Rolls Royce procedure is based on the more definable value of "i life to bu;s-t" obtained from spin-pit tests,

Of the five topics highlighted in Dr Snide's presentation it was one dealing with the Allison electrophoretic coating process that attracted discussion.

In reply to Ir. Moin it was said that the electrophoretic process couid be used for applying internal coatings provided that the blade cooling passages were not too small. The major advantage of the process. however, is that it gives a very uniform coating. The thickness after diffusion is typically 0.003 in. This led Ir. Stroobach to raise the important question of the measurement of residual coating thickness after engine service. There appears to be no adequate nondestructive test and the only acceptable method involves cut-up for metallurgical examination.

Several participants (Or.Momn, Dr Worth. M. Mazars) asked for more details of the coatings and for comparisons with other coating systems. The coating for nickel-base materials. AEP 32, is quoted in the paper as Al-r-Mn and is a diffusion-type coating similar to plain aluminide coatings. It is deposited in a propanol + nitromethane bath: details of the bath composition could be made available if required. Test results are given in the paper showing some improvement of AEP 32 over standard aluminide coatings but these results were obtained from test pins and not from engine experience.

IMr Plumb's description of the repair methods used for helicopter engines and naval propulsion gas turbines led NMrDeutsch to ask about the causes of deterioration. particularly the importance of the higher aromatic content of present fuels, a possible higher sulphur content and effects due to erosion by incandescent carbon particles.

In reply it was stated that there are as yet no problems with cero-quality fuels used for helicopters The diesel fuel used for nasal propulsion is rather variable depending on its source, and both higher sulphtur anti higher vanadium ,-ontents are causing some concern- rests are b'ing done in UK Governn , : Establishments to determine the effects of simulated lower-quality fuels. Carbon erosion, not necessarily from incandescent particles, can he ser_, serious and can only be countered by improving combustor design to reduce carbon formation.

ThI presentino 1,i'tven hv I r Stroobach drew attention It) managing operational and ntaintenance proccldurcs in order to minimise costs. The main factor is fuel cost: materials costs are small in cnomparison.

Professor Iloeppoer asked for more background information on the engine de-rating schedules that are uscei In reply it was said that engines are de-rated as much as possible. the maximum at present being 1S':. le-rating at ta;ke-oftf has the largest impact on ergine life. Flight crews are unwilling to reduce the take-off thrust below the thrust uod duhring

IiLmb and this consideration lirnits theamOInt of de -rating that can he used.

-A RI-2

The USA programme on Retirement for Cause was described by Mr Harriý A eio~idemrhle apwvsno of development work. particularly connected with non-destructive evaluation (NDIV. is required in order to meet the target date for implementation of the svsteln in I 985 The progranmme was reported to, be on schedule.

Itr Morn drew attention to a recent proposal that first stage fan discs in the r-10 engineg ihuld bQ given a proof Test and eddyv current inspection after a given numiber of cycles (pierhaps 18001 and then returned it) %er%ice. Thij. inspect. return sequence. however, would only be permitted for a small number of times 4perhaps twice) and theu, retirement would be mandatory. Iln contrast, the Retirement for Cause procedure envisaged an indefinite number tof inspections and roturtns-to- sersice. Did this inmply a lack of confidence in the present procedures? In reply it was pointed out that 11Q85 technology -A ill be an appreciable advance on that of today and it is necessary to proceed cautiously. Furthler. although proof te-ting and inspection may confirm that a disc contains no defect likely to cause burst, there are other lire-limiting factors such as pitting corrosion or fretting at dovetails which must be taken into account. It is not intended to change any wear or dimensional limits in the Retirement for Cause procedure.

MrI Jeal asked about the basis for selecting the defect size at which a disc would be rejected- The answer was that two approaches would be used, one being the normal fracture mechanics Calculation to predict a defect size which ýoultd not lead to failire during the next inspection interval, and the other being dependent on the outcome of current work On NDE. Regarding the latter, a probabilistic simulator is being developed which will accept information on detectability and reli'ibility of available NDI: techniques and will then be used to select a limiting defect size taking into account appropriate risk and cost criteria.

Professor liceppner returned to the question of the criterion for crack initiation; whether this would he taken as "'first crack" and whether the saine criterion could be used for internally initiated cracks. fie also wished to know how the current NDE capability related to the fatigue lifing methodology. In response it was emphasised that the characteristics of the material being used had to be known in verv great detail. A large number of discs in INNI 00 have been cut uin for examination and there is now detailed knowledge available concerning the statistics of the siues and numbers of defects The Monte Carlo (probabilistic) simulation can titus bie used to predict the likely Ivehaviour. For information about internal defects several different NDE methods are used. One of these (eddy current I smihld reveal just-subsurface defects within about 0.2 in. of the surface. Experience with the inspection of earlier engine discs is not a good basis for developing Retirement for Cause NDE methods since previous engines incorporated less sophisticated designs and ran at lower stresses tinder less arduous conditions. Detailed inspection of new discs. on the other hand. would bc useful. they would be "'known'" as a result of periodic safety inspections done before the critical inspection on which the decision whether to retire or not is based.

In reply to a further qjuestion from Professor lloepprier it wa~ssid that thle Vexpected ads dilages of a Retiremlent for Cause system would be realised in temis of reduced "osts by extending component lives An alternative advantage could be to increase safety but the present policy is simply not to increase the present risk.

\Mr Jeal outlined the more general philosophy being developed in the UK oil componcnt lifing :itd emphasised tlhe considerable amount of detailed information. obtained under realistic conditions. that is rCquired.I t woutld proide a basis for On Condition L.ifing where this could be shown to be economically attractive.

Professor Wanhill wanted tonknow tile differences between Retirement for Cause and On Condition l ihing and wýas told that they are essentially the same.

The more general approach. however, is as much concerned with the first prediction of initial life as with the possibility of life extension by periodic inspection. Both demand a detailed kno\%ledge of the oc,.currence Of defects and their behaviour on a probabilistic basis. I5.n. !ariJ nirs en

-nlustrieterrein Spikweien 4 5 3 AC LOM" The C•etherlaInds

ri-h temperature coorrcsien in irniustr•-la and marine gos t:.rti.nes is the result of the in.Žreased inlet rae terperature cf the t urbin e, whi1- : r.e the ecnony , and the use of cheaper ruels, which1o-ntain mere u In order to reduce hi& te7-perature aorr's''. Ti-Si ai was developed on the laboratory scale by tests have bee perfor' erd 2'.' fte" dc'erren f t 'oat I n<, sndr r better understarcdin, the benavl"ur a-h rat' 4 t , into teorsi t sechanis-s is r.iuired. A 7-.1 '-' I's -- en b-! -t'' ), he b-.-s f the laboratory ..Jevýioprent. T'-"--- , teninenurther 'evelcpedi to [rr 1e an industrial cotatt a Dcx , for usE n hot norponents in

2. CCAT'<. vi.ff'.C70P ANC-

The first step in ý ie ,ti r • ?.. • tk - th .in t'-:ni'- layer on the sur-4ce. "One of the best methods for thns is ion polatinn 4. n this process the surface of' the coSmponents is first cleaned by a glow discharge under low aron pressure. The titanium. is then evaporated while the.' 'h- biras voltage remains. This gives a pure titanium layer Sof' t to 1 microns thickness with ve ood adherence to the base material (see fi. ). S"Inthe second step, the titanium is diffused into the base material by a hewt treatment in vacuum. As a result of this process, a surface layer Of Ni Ti i' formed with a thickness of abet S30 microns (see fib. L I' -. the lasts Ste the par s are " ci by a pack process 9). in this process the parts are packed in boxes in a mixture of silicon powder, activators such as chlorides and filor des and inert aluinium oxide as ? diluent. These sacked coat boxes are heated for 1.5 hours It about 9001C. Durini! thbs process silicon is transported from the pack to Lhe surface and reacts with the Ni 3 Ti layer at the su/rface. A very complex layer is formed in which G-phase, '1" -Phase and Niq Si is formed (see fig. Ic). Fi'. 2 shows the chang'e In concentration 6f the main elements in the cýi, Contrary to most Of the :,thir vratiorr esses, the heat treaitmient of the base 'ater is not carried out after 'mcatin fno mat- n, but is combined wi thhe ý'atine p.-duýti.n, The diftusion trea""-ent 'or the titan-um -is conbined' with the solution treatment of the base materia'. After the licoorn at Q00"C the base material is ared, The de'os'ticn of Silicon at "c causes h toe •-asi prime to tend to ""era.e a little.e. !wever," ,r IImost of the n, -elbare m-datleifference in 'a-drles is found .p wit" the optimally heatrea""mtei- The reason why tin- heat nttci' nn, he r '"--'- f.r silicon n, t Scaiin; 'relt.Oil 't te--mratur-ts :I l:ut -in! i. i te<:pera'ures (r:'' ? t

.sn lrt i i, tn1 -.;::e. - ''A,".

.IIx • i 11A , i . ai t. t. Ax` in-han~ fl, ]nt r~l•. _'r•'..I.r

A nto 7yr - r-

Pr.-. C :: n 7 Inc , ": :. n -• .

- ,, -. l...... - "'. . h 1%. - - " Y A "" a - *:,.e- A !Vit - ,. 7;'--] - "-.-f:: :.. ;; ; P 1- . -t" -1 L'"; ": ... , -?' - " :.. 1 " -:. .;' s [ ' :. ": : y,

"R:u:o• W-1.:'• ZhSne - ".7-. i.n!._ W=,; -1n ..o V"x ..- .' ' 7 ý.; 1: -%n! i-"( Z: I- . n

S t, ir'.e s in .

Afthuinu h the rio -f n-' -nian. %n -no' .' 7%o- 1i z .... n . t

'Qdil%0 xdation'sr i-• a'" at !'-'',7 ' -s "'% 'ecc--",-..-"'-.•~ - .s,-i DifustionE -'at 360 u'tt tem*permture.sif - 'on irr o-ver intthe':.-". b-ea'ePrial- -ii "' *"- ,.'a-'31-

t flu-"" t 4'-e . t e e..,nt

in. ae-Ph fnothe"-aton -inthe -'A-'"-n" i -'-f'3 n in - nn!; X l -'v

For Qoed stabiliv'-.t c , 2 '.•t t'-:'e t :, 3'''•• V] ^[ . . "Ir.'•-t'.:- - •' • doeoC is necessary. In practicenhis stabilTzationPccurs iurtnr nornal tn -fto - operation. The concentration distribution of the elements of Elcoat 360 an in "73$ directly after coating production is ,aiven in Fis. . 30 hours at nAfter550"C the concentrations ar e cn fg. 7). The coating is then stabilized and no further structural h-anges take place over a loner time.

m4. CONCLUSION

It has been shown that the titanium silicon coating El;oot 360 offers goca corrosion resistance against low temperature sulphidatien and hot forrosal. m 4.COCUSO5.• ACKNOWLEDGEMENT The author wishes to thank N. Swindells of the University of Livernool for his efforts Ithsbemhwin the X-Ray micro-analyzerhttettnu work. iioncaigEca 6fesri orso

reitneaantlwtmeaue upiainadhtcroin 5.AKOLEGMNThuhrwse otakNwneioh nvriyo iepooi

efot inX-a the mir-naye work.I 1I 1l1 m m~m• Ay 2sC ,•iC" : Ž-,.l ato .. for' In 7A

,' n 1 eoport, Cost.,-) - ':2 p1o,-'0 I: et-- 10 6.

",,t,c,,crr'on :c'n : as turbines" •C,:ipo"rt 2-- June -Iy2.

co" ; *;.!." tht h 'a e:'rto, : .of uin-oat l.cned TiS-i c.oated Tn 72,ý !.C in a

ii"1 S-I ib .'1.I - i ApI' i 1 .

New `Yor'k , ]ii•

:• .N "wiJnd-' '- SIllP~l•." hoA • r'I - 'F[l i ' "{ ' vtl:• j cilI[..tllii rCO;-itjii&''

ii' oo-7

S. . ., I i..H i i* i ' l .. . .1 14a

A ,.. 5

S.9.. ' l

V.. I

i 7-5 . 7

r 30- "II ' ' ' ' • ' . ' ',.

I "I I I \ ---- - i------AI

I ' ' ' " I t ' 'I I

10 urn base material coating

Fig. 2 Distribution of the elements over the coatino Elcoat 360 on In 738 LC base material.

suilphidat on hot oxidcfl ion

corrosion

700 80) 900 1000 1MOOl

so temperature in 'C

rig. 3 Possible corrosion mechanisms which can occur in a gas turbine as a function of teosoeraturn, III -v •, ,•';

J•.~

gbi 4 7 - A t ?y .#"- a '744' ,J' *:t'w V - ISDx

h- 3 -,tc 3

uP !Vlr.1 ZL I t 75 ii' - Pr? - I M 'I A5MsV - GOOx

pwI

-Vt dZA

TIN 1 ut .11 ii urn ti wi' it I g -ri S'.8

30-

20-

Nm 30" 60- /.

40-I

I -• 20 10 /\/

2- 0 1 ,"- , " \.•

-Cr

I ' I I' I I" l I , '• A

base 10o uni material coatinq

Fig. 6 Distribution of the elements over Elcoat 360 on In 738 LC after 300 hours 850"C in air. i " 8-I

INFLUENCE DES TRAITEDQ2TS DE PROTECTION

SUR LES DROPRIETES MECANIQUES DES

PIECES EN SUPERALLIAGE

Par HAUSER J.M, DURET C, PICHOIR R,

OFFICE NATIONAL W'ETUDES ET DE PECHERCHES AEROSPATIALES (ONEIA)

92320 CHATILLON PRANCE

1. NTRODUCTION

Les conditions d'environnement des pilices constituant les parties chaudes dcis turbanachTnes adronautiques imposent le plus souvent la rdalisation de revetements protecteurs. Ceox-ci sont d'autant plus n~cessaires qe la comnposition des scperalliaqes utilises et le traitement thermique assuci sgornt choisis pour obtenir lesmeille•ures propri~tds meanivues 3 chaud. Ce choix de coripositiot ne corresponid pas, en gdnmral, a une bonne tenuo a i'osyuation et I la corrosion ; quant au traitement thermnique, il nest qgneralement pus cunpatible avec lo, cycle thermique de realisataon dIes revtoemernts protecteirs et cet aspect est rarement pris en coaipte par les 61aborateurs et mnime par les bureaux drtudes. Pourtant, cycle thormique do protection, caractdristiqges m6caniques des materisux 'tonstituant les revetements. interdiffusion de ceux-ci avec le superalliage, peuvent avuir une influence difavcrable cur lee proprit:tes micaniqueu des matdriaux proteges . comporternent en fluase, fatigue, fatique thersimpe ...

La durre de vie des pi~ces nest pas seuleruint dtex-minCe par les propri6t6s mrcaniqur-s do sureralliage protqg trais &qgalement par ia tenue du rcvttement d loxydation et A la corrosion rennou- ve er la protection peut poser divers probl'mes et, dans certamns cas, constituer un nouveau facteur de rexý.ction de dorCe de vs'.

Ces divers aspects seront d6crits plus et ditaii cit illustri's par des r6sultats concernant dcs revetesents d'aiiminiores ap~pliqti~s ax soperalliaqes IN I0' et IN 73HCC. Avant d'expocer les, resultats obtenus sur lees lliages revktus, nous avorts voulu mettre en 6vidence l'.,nfluence d'un certain hombre de facteurs sur les proupriatc-s m6caniques des superalliages, factours ir~htr.'nt s a 1 'tlaboration -le la protection.

2. EFPET DU CYCLE TI1ERMrIQE AStOCIE A L'OPERATION DE PROTECTION

Los upi~raticns du protection, qoi eIrccxlnrtenitle plus souvs'nt ao co~ins oin meintistn plus no mincus prolongS a temp6rature moVenxe ouGe evee, ont gdndralement on effet sur la rincrrstructuru des iýuopralliaq.s, en particulier sur la precipitatlon do la phase y-' LNil (A1,Ti7l des alliayes base nickel ou sur la nature des carbures iziteryranulaires, -euvent intervenir la doir6e CI la temptrature du palter de tritem--nt, t ""yalement, dans certains -as, la vitesse de refroidisrement qui reste tributaine de la misc on ,.oeuvre! *I proct'-sl dc protct ion.

Par consequent, le traittiment de protection peut affecter leB prtpxiri-tes mi6c'que dex. l'alliiage non seulement duns la oine superficielle .ais egaloment duns trit IL volu'Ime. Ce i esC't iei'o'ot

mrs eni (vidensce en soumettant de(. 4Obauches d('prouvettes auo traitemrients d5 proteuction, M. en r.imutant eýncoitop par usiriaqe l~a torie 'pr olyaffootIhs. chitriqoexnent par lo trait E,(n4.e

Dans It ca des rx.vetcminents tallia4e tIv typer r-Cr-Al-Yal-• '!; jar x li 1--", -i pressict r.i oite, jar .-x,.oratsin ther-ni ueŽ 'u par d1, techfiiqi--'.t q ri Ive. •-i .n-i .ti ,itt l--i I t'Xitit al- pri- "J au ffaj ' (il:es i ý,ic et temti,A p itr re m i i'o [o , ,er r d r t ; i t'.es t 'i-.5 i , p , a - . -i'. r I .A t-. ie- ti c e Iu dej-A t l 6ten i ,tilt.utd a ris 1 '-I omt , •t ,T. i i--i -' - . uc - . i ], ;, li ,i - hi-c o re ,i un tr-I c','l It ther ni-t• ui t .l ,'- fet•1u6"' . i'i isis ;f. i' s.1 '.I doj I i-iti ri-I i i "-''; 0I di' pt:.-cipitati,'xt do la t;.u.z' (aI' 'ialu es I',.e.-tucks-l,. , il it' -. • is 'I•I;tt• it1 .a-[p'-riso;i-i' .l •'!

~ri•'t.'-s mncuttitjucs e i.'i,•ýlll t . II'tt1h0 ti, si• t., P , -- tt;I. 1 ,-i,,i.i, j ratio- S,•,•6lz.•ant-i~-rer I' actcrichap'J .o I ,l.-. -t .,,i':etit '.jaieiti.-x rester .7-rift..' ii-l.'s •Lix 'r ii" ru .-etl~o to!l-r,'t:41', I',, a';i',..On Ivut, jii'i' 'ti- hi's' . i ' " .i 'O 'iii 1 it" "

l',iLiag;e Ialliaq~s bas't.s nit4c-:I

'saxti ixIu c',, '''Is tlA-5-- i.;s r cnii 'Itn '•ti ri' to x- i sti; i .IM.-xe'1x11-1 i it.- 1 i

-I - - i't " ,I. II :I- -. 1-,-I' ,i ij .0 i, i ti' I Lt 1 In r -I.t

'In, x it•it .18-2

et mains fragile (par exemple en 5 h a I 0801C pour obtenir 60 um de Ni2Al3 sur IN 100). :1 est souvent possible 'e faire coincider ce dernier traitement avec le traitement de mise en solution de la phase P dans les alliaqes base nickel.

- Les rev~tements obtenus par aluminisatian basse activit6, parfois plus performai•ts du point de vue resistance A la corrosion, ont une cinetique de croissance plus lente, contrl~e par la diffusion du nickel A travers la phase g-NiAI riche en nickel (60 us de NiAI sur IN 100 en 16 h a I 050'c ou 60 11 A 9501C)_ L'incidence de tels traitements sur la morphologic des prdcipites de phase I' clans les allages A base nickel, et par cons6quent, sur leurs proprietes mecaniques, est qgneralement importants. De plus, la mise en oeuvre de refroidissement rapide A l'issue des traitements effectu~s en "ack" est difficile.

- Les traicements de chromisation, qu'il peut Ctre int~ressant de r4aliser avant aluminisation des superalliages, sont toujo-Ars effectu~s & temperature sup~rieure A I 0O00C (30 1 & 1 050WC ou 5 h A 1 150*C) et posent les msmes probltmes de refroidissement que prdct•dument s'ils sont effectues en "pack".

A titre d'illustration, l'influence de cycles thermiques d'aluminmisation sur lea caract6risti- ques de fluage A 850-C de 1'IN 100 et de VIN 738LC, non prot4ges, est presentee su' lea tableaux I et II [Il. On note que cette influence peut varier selon la duree des essais ! des essais longs conduisent parfois & une amelioration des durtes de vie des 6prouvettes trait6es, A 1 inverse des essais courts (tableau 1, lignes I et 2). Il est probable que, dans le cas de VIN 100, la microstructure obtenue aprbs t-raitement A I 050-C n'est pas stable et est susceptible d'6voluer lors des maintiens a o501Cen fluage lent, vers one pr~cipitation de y' plus favorable A la tenue au fluage. Pour dvaluer convenablerent I'incidence des cycles thermiques de protection sur les propridtos mecaniques des superalliages, il apparalt done nessaire de rdaliser des essaim Ca fluaga da CurC-e muffisamment longue at de ne pas me limiter & dam ammais de courte durde.

3.CONSOMNATION DE MATIERE LOinS DE LA FOENATION DTi REVETENEN¶

Lors de la formation des rev~tements par aluminisation., le supe•ralliage est transformi, par diffusion, sur une certaine profondeur qui depend de ia nature du traitement et de la crmposition du substrat. Pour les pieces aluminisaes, la section utile de superalliage est done plus faible que la section avant altuminisation. Ceci peut avoir un effet non n~gligeable sur Is tenue en flirage des pi~ces A patois minces : nous avons en effet montre que les revtaements d'aluminiures ne contribuent pas 6 supporter la charge de fluage (voir plus loin § 6). Par exemple, Ia formation d'un rewstement d'aluminiure de 60 Lo d'4paisseur entralne, sur IN 100, une consuomation d'alliage d'environ 40 im ; sinsi, pour une paroi mince d'epaissenr initials I am, 1'dpaisseur dIN 100 restante aprls aluminisation de cheque face est de 0,92 mm. L'augmentation de contrainte correspoardante (X 1,087) c(harge constante correspond A une durde de vie rdduite d'un facteur voisin de 2 (500 h au lieu de 1 000 h a 850oC). Dans le c-as d'une paroi I'Cpaisseur initiale 0,5 mm, la reduction de d'rr~e de vie serait d'ul facteur 3 environ, a charge constante 6qalement-

Dans le cas de l'4laboration de rev~tements par d6pbt d'alliage (P.V.D, projection plasma .. 1, colle-ci nimplique pas nne conscummation de superalliage mais entraine seulement une interdiffusion substrat-revCtemeiat limitee. Toutefois, le revtatement d'alliage constLtue alors une surcharge (masse supplimentaire soumise a lacceldration centrifuge dans le cas des aubes mobiles) supportie pour l'essen- tiel par le substrat de superalliage, ceci dans la mesure oil Ion admet, creme pour ils revdteaentsd'aluminiures, que le revAtement ne contribue pas A supporter Is charge de fluage.

Finalement. pour les deux types de revCtements, les rapports de la section tctale des pi'ces determinant Ia charge due A l'acc~ldration centrifuge a la section de superalliage supportant effectivement ce'te charge soot sensiblemeiit comparables.

4. INTERDIFFUSION EN SERVICE ENTRE LE REVETEMENT FT L'ALLITAE

Lors de maintiens en service A haute temperature, de nouvelles phases peuvent se develepper par interdiffusion entre le revhtement et le substrat. Ces phases peuvent eventuelleme;zt jouer un r6le sptci- fique dans la tenue mdcanique des pieces, selon leur isorpholnqie (couche continue, pr6cipiteds dans le revC- tement, dams lalliage sous-jacent au rovAteaent), et selon leurs proprietes (ductilit6, frigilito). "tte morphologie d~pend essentieltement de la nature du substrat et du mode d'6baboration du re-vstement (rev,- tement d'aluminiure obtenu par diffusion, revtement obtenu par dCp6t d'alliase). Dans le premier cam. on obtient gqn~ralement a Is fois des pr6cipitations inter et intragranulaires et dCs ci'chi-s continues de phases intermddiaires, alors que dans le deuxiOme cas, par un choix approprii( de la comjapsition de l'allicae du dip6t, 1'interdiffusion peut Atre limit6e A cse variation progressive ie canpusation sans apparition de phases "fragiles".

4,1. Intfrdiffusion entre un revitement daluminiure et le suueralliaqe

Les revttements d'alwniniure, formeds A la fois iar apport c'aluninium p'ar voie aze,-Use et p.ar transformation de Is zone superficielle du suoerslliacJe, n peýivent met ire en solut jr-n qo'une faible payrt n des 616ments c'addition de ce dernier, Clementsa ui prOcipitent Aans la zone, interre sc-us ftere I 'arlurns ou phases inlersdiaires (', tar exm .ple) riches an Cr, Mn, W, Co ... ? ,Tltte-nulnc. A liffuorr dans 1.a zone .;rous jacunt, du bubsi rat ari -)urs +, mais tiens in tem;.jrrat ur" Selar l'alliage et la temperature de mainrtien, onr note i'apparitiun cicuc!t! rnopnas6es i t et/ou des precipitations in.tergranulaires (car-ores) OU intrnqranulaires (0, carburos ..- ) sir une pno-

foder .Ai~, asIccsdo1 .N1el, -i-'.'"""-AC) ziusrv e ni~cjr~tac ions - intergranulaires de carkbucs de chrome sur 150 m sous le revyteaoLe ajres 3300 h do maiitcen As lk• (figure 1a), ou sur 250 in sous le revItereent apr'2i 3300 h A Ii ,CC (fih h et AAcI, aiisi qu'utie prt-cipitation intragranulaire de phase aciculairic 3 Isur 2-1 x. aprls 300 h A 850'C) ou di carbures de chxaore (sur 50 cm aprts 300 hA 1 050WC).

Dans le was du fluage, il a ett ubservi [1] 12] clue dcIs fissures as'orc~es cans le rav~rexecnt .. un stade avanc6 de fluage peuvent se propager ae tow, des plauqet~es ,a 'on, - dans 1'alliaye tfiqure 21. WNanmoins, il n'a jamais 0t6 observW de ruptures J- p-artir de telles fissures, ces ruuptures ayant touýoursO A 'origine des d6cohdsions inteairanulairesrc-nbrauses dans I'IN I0O soumis au fluage [L1- et cec: m.csm'aO voisinage imucdiat de fissures se propageant sur ins plaquettes de phase a intraqranulaire's (figure 1). Les pre-cicitations intergranulaires de carbures dC chroM.e dans lo substrat au voisinaqe dCI rev6tement V'aluminiure pourraient, par contre, darts cea conditions, dvoir otno influence sur le comportemeo,,t ec fluaqe (ligure 4 c observatior d'unc dCcohwsi.iin Cal'interface carbure, grain y - y.,) 11 faut ajouter quce dacs le cas ou sa forment des couches monophas6es eta 1', celles-ci ha contribuent par A- supporter id charge -dea fluage.

Daihs le cas des scYlicitations de fatijue., la pr6oipitation acicula-re c3k ,.hase c pourrai" avoir un effet plus dtterminant : er. effet, il a 6tc$ montrcj [31 que, .q rs des recoits do : 0 iiC CA8-1- C, 10NCO 738,LC aluminis6 par le procci.c6c LDC2 prOsente unOcbaisse de durot J- vie on fat rse vilcratzire A 850C que I'on peut ccrr6lcr avec on anqorgaqe de fissures dans I.a 2-'on, Ie rsv;-iptit on deia phesv

4.2. Interdiffusion entrtaun rev;tement demc*(Valli Lie

L'enterdiffusion entre la revsotement de type .ICrAIY et I' sabstrat peut Otre tris recioite car, d'une part 1a diffusion es'. limit6e pendant le processus cid di,6p6t, et d'as'-c. part un choix ac-irlrorie doe I nuance de revtirement ic*rmnt d'cbtanir rapidement unes boone stabslit- -'ies phases en prisenc'e.

Dans certains cas, wependant, 1' interact con peUt ýýre uspoýrtante ; o t exreMple, 'lass It -- s deC revtements de type CoCrAtY dtipumss sur un suporalliaqe base nickel, la fraction voluhcique 6 -a phase riche en aluuium diminue proqressiverment, tandis 'rue se >ronne uone wouche interm6diaire Ile _ ssC * Ni IAI. Do tel1es couches d'icterdiffusion ne prdsentent pa- de frac ilitlc marticoli'tais ne ocoictru.juent I-a A supporter la charge de fluage.

5. PROPC0k•h M DFS PIMArS C70NSTITA'ANT LE RVE'TijAA:PTT

Les revdtemients obtenus par alomini.,ation coiprennent .j6nrale:n-t one mttrce ',cluoini.cui - I( - NjAI) et des prtiwipitw's ; leg rev~tements d'aliiage sont constitu~s cI phase i - s'c base Co oc Ni, riche en Cr. g6noralearent pr#ponddrante, et de phase C- - NiAI (ou Co All . te cartccýfre fracqcle 3 Las',a tem-6ratare des intermdtalliques NiA.! et CoAl d6termine le ccccsportement de.a rovvtaccnts daars lem deox cas dans les revtrements d'alliage, IL caracct6re fragil, ost attinuci par la pllstieicil,?i. laU s (cii constitue, en gfndral, la matrice du revuteient.

La tempe-rature de transition duct ile-fraqile 6c revvtCements 1'alkLomioiure a 4t6 6tudiOe, en partiuwoIior en tract ion a1 'aide do mC.hodes d'ofnission acoustique 141. Pour la phase NiAI, rlthe en alscixciui, elle a Cei ldr'crdre de G5a) 3 750C I au-dessouo de 650•C, lea casskires ajjparai-Isant -2n tracct ic-i pour des A1lnqations de C0,5 ! 1 ,51 ; au-delA da 75;'"C, lea reviteiments d'alanniniure JeUV.cnt otrtau contraire plus ductiles qu le. suparalliacie at pevont subir den allonateiacts importants sans fiszourker

(ti-3 rc 5). Les aluminiuros de cobtlt apparaissent comic plus fraqiles .otu les atuminicres de? nickel ,1A]unqcsm•i't •A rupture en trartin 1- '3l'"tirve do O,25 4 au-d:-ssous do "

t 0 Les propri6ts mscani.'u1 os des r ci "'son doalli-aqe d6 1 ''ch'nt des cp- jctt s cod'clot 11c .cc icoux Ilasme- c'oosttt ices ainsi -ue dI loars pr.orticono et d, leul cmuorhl-.clcic. lDans 10 's . Lcina ieos b i has(',s ,/,"- (Co-CcAlI ) , re r istL_ I!' ms co,t rIicl-st ata ,n d cucomp, t cct miu-cn i ' ue l' i : n -

,. pv; etrr 'lunri rv- h citutc i isa n c eI _ i-.s¼- ) " uciI a )' cO1I r eask' -- l'sta;c )cu. i :natrie c: cIcc.'mcj.cccrt ment cchin- I 'cr:.-quc' 1h, 1-h,--s acc'-.-- ''tot i c'' - c -crc I-- cr-co--c i''cc-c1 cli-;-asn" .:P- 1, u n ILI'-c:cI"an d''m cc~dcc sc- -c v 1-s 1!a 1 AlU -, itc-i 5-I'-' c--i. v ccc-t Ic 1 coia a -, 'it T ' IZ Ic ,a ' vr cIt ,ici 'Act cioan, I, in 1f ti1 5

ý1ir',",-,. ,acc- jt- -

I h, : [zl;[ £ I [£'F ME•'AN•I &U[EZ 1"::r i MtCA%, EN[ -iUNA[:}.A1.1:A',L: i Ifýa cIt III-s

I., cccc - iu -(- Iccc'-} ' A' ccci;[• :,••- 12l ci.;x 1icc-, 0; "€: Ic ccc c A U A"I( ' ' :'( -- tIc• [ 11I'I

I '~ - , - ' ccjj

-I'I '- -- 11 III-I I I 1•1 8-4

,-. Los essais d* longue luxr4e 121 (quelques dizairns do milliers dhneures) d ESýCC sur T10CC7181C e.*a FIX 4C- aLunnis(--, ayant subi le traitement thermique standard apras aluminisation, montrent que la prtser''a du revftemrcct est sans effpt sur la durie de vie -La conraintae_tant calcule Sur la section de j*ýpS~ytIteavant alrssinieut ion-.

Lea essais. de ccurte aL noyenne dur6e A 850'c sur IN 100 ea INCO 738LC aluminisos IIl montzent quo ce n'est pas I& pVrisanca du rev~tement en elle-snme qui modifie la durde dt vie du fluage, le temps d'allouceenmant a 11 et la vitasse de fluage secondaire (figure 7) - la contrainte 6tant calculae sur la section r~siduelle de- suverallia'ge apc~s aluminis~tion -. En affat, La casparaison d'essais sur Oprou- v~ttas rues avant 1)cotction, protdqyes, nues mais ayant subi le cycle thermique de protection, met en Ovidence que la degradation des propri~t.is macaniques observqec clans les deux derniers cas pour les essais courts, doit Etre imput.•e au cycle thei.ivrae cde protection et non au revc-tement lui-mame (cf. §2).

Ces 3eax Ltudes tenderet donc a montrer qua Ia presence du revitement est sans effet sur la dure de vie en flhaga.

Le fait de le pas calcoenr la centrainte do la mni.e fagon dans lea deux cas ne modifie en rien la conclusion prddddante car, pour les Oprouvettes utilismes -6prouvettes cylindriques de quelques milli- mtres de dua-nitre-, les correct'ona de ontrainte sent faiblas et leur effe!t sur la dur-e de vie s'ins- crit sensibiement dens la disper sion des rdsuttat.. IL a AstL toutefois montr6 par des essais multiples [I] que la revAetment d'alomiai-cre subit la dAformat:.on nmpesde par ia partie ctntrale en superalliage sans porter une part appr~ciable de la charge at, de ce fait, ii teste prdferable de faire la correction de contrainte ; par contrea, cette correction est indispensable dens le cas des 6pruuvettes A patois minces.

L'observation des rervtements d'altusi:iiures apris fluage A SSOWC ill mont-'e qua ceux-ci prO- sentent une plasticit6 Lmcsjrtante en fluage A carte temperature . A 4% de deformation. lea rev6tsments sent encore exempts de fissures ; des deformations encore plus importantes sent ioYssibles localement, en particulier au droit des cavit~s intergranulaires qui s'ouorent dans le supetalliage (figure 5). AprOs une deformation globale importante, des fissures apparaissent daens le revftement so nrvaau des zones de striction. Pratiquemont, de telles fissures re soret ja•.ais ranccsstzce.s n'r oil-zen rfellas, ru'-sque aelles-ci sont retiries du service apr-s I A 2% de ddfonsatin',a : teur6IVc 6aentuel sur la rupture a cependant etd discutC (§ 4), puisque leur propagation a 6tO obr*itvde rur quelques dizaines de ýa dans le superalliage le long des plaqoettes de phase o de la zone d'interditfxion et tI a 6tO montrA qua, dans nos conditions d'essais, ces fissures ne semblent pas initier la ruprv P qui rtste intergranulaire dans le superalliage (figure 8). IL parait important de rappe..er qua, par contre, 6tant donnA la nature de cette rupture, Jas phases (carbures) ddveloppileo par interdiffusion rev~iement/substrat dans lea joints de grains de cc dernier peuvent influencei le caeportement en fluaqe, en particulier dans le cas des pieces A patois minces.

La rdsistance A la fatigue eat ine proprictk particulinrement sensible A L'c'tat de surface et a 1environne.ent, puisque i'amorqage des fissures a neuvent lieu % la surface des pifces. Les rev~tements peuvent donc modifier de fagon niotalAe la tenue de ces derni4res si leur rdsostai:.-e chimique A l'environnement permet d'dviter les interactions corrosion/fatigue ou oxydation/fatigue a'cxquelles pout Etre sensible le superalliage. IL en eLt de msme s'ails prtsentent une fragilit6 riarticuliere dans certains domaines da tsmpdrature ou pour certains types de sollicitations de fatigue.

-•Ler61e spfcifique oAure rstement, en fonction de sa nature et de la teinpdrature, a seulement pu Atre mis en Avidence sur des mat~riaux tels iLs cuperalliages solidifids unidirectior~nellement poly ou monocristallins (example S 200 [61), lea alliages torgds (UDIMET 700 [6]) ou lea superalliages composites a fibre do =arbure obtenus par solidification unidirectio.inelle [71, alliages moins sensibles A la fatigue qua lea alliages coul~s Aquiaxes. En effet, dans le cas des alliages coolds equiaxes, cc sent des amas de prcasitcs int-ernes qui provoquent gdndralement l'amorsage de la rupture (figures 9.10.11). ce qui rend les revftements sans eafet sur la durde de vie en fatigue (850'C, grand nousbre de cycles, figure 12).

Saloe. le type de sollicitations de fatigue, il faut distinguea : la fatigue A grand nmcbre de cycles A la rupture (HC?) et par consqouent faible tauw tedo formation, la fatigue A fort tauo de d~formation et par suite faible combre de cycles a rupture (irF). Dens tous lem cas envisagda ci-aprts, la u6for- .ation du revlemenrt est imposde par cella du substrat.

atiefaibla taos dea dilfutmiation

La deforsation du substrat de s'-.eralliaeg est alors essentiellement Alastique et, d'une manidre gdndrale, les matdriaux constituant le recawtacent auront tendancea A voir une bonne tenue en fatigue s'ils ont une limite A1astique Olevee ou s'ils ont un faible module d'61asticitd. Dans ce dernier cas, en effet, pour un tauo de d~formation impoma, pour deux matdriaux dcnt le module d'Olasticitd difflre, le rapport des contraintes qu'ils supportent est fgal au rapport de leur module d'.lasticit6 respectifs.

Ainsi on revetement d'alliage Ni-20Cr-IOAl-2Hf-O,IC sur NiTaC 3 - 116 A2 (eutectique orientO contenont des fibres de TaC) a uri meilleure tenue en fatitue qu'un revdtament d'alliase Ni-2fcr [71 dunt ia limite Olastiquc est vraisembiabije.•amnt plot faibIc, - iu•i peut mkne conduire a cv qgo' l soiltsollicit6 dans le damaine plastique. De meme, a faible taux de ddformation, on1ravCttment mq-se fragile pourvu qu'i soit soilicitl en leqgA de son domaice de ddfonmation plastique n'a pas 6ci Ll' •"--.. sur la tenue en fati- --ue c I-ci example, il a 6tfl AservO 161 qua l'aluminrsation dp I',IDIMET 7Q0 -acciviv de polissage Alectro- lytique pour annuler tout eofit -3e concentra'ior de ccntrainte- no~n seule:ent nae c'-týriore pas mais t ,Am liorlu le ccoaporttment eln fatluue A 1'.icbiantv dc, IP''DIJ.f-rT 30.( nro r ycSte Ot 61%ler'n IA i. lnvars .ý •.t, lfrsqu.' la r v-v6temernt scib t one d6fc•crat con plc st icjue slurs r |rrc Ic, subst rdt t,;% (enoý"t dans le rt-isatinc 1 '|t i ;-| - Dans ce cas, le substrat et ie rev~tement sont sollicit~s au-dell de lear limite 6lastique.

Si le revgtement est fragile (aimiiiure. Alliaqes y-'t & basse temperature), dts fissures peuvent. y apparaitre dds le premier cycle et r6duire notablement la duree do vie en fatigue. Ainsi, La dur6e de vie de UDIMET 700 est rdduite de 20% A 7609C lorsqu'il est revZtu d'un aluminiure [61.

Au contraire, si le revetenent est coctile (aluminiure ou alliages y-," a hautes temp6ratures, alliage Ni-20Cr), la tenue en fatigue Lu superalliage nest pas d6grad6e et peut mnme Otre amelior6e. Aar exemple, dans le cas du NiTaC 3 - 116 A2, altiage danu lequel it a dt6 observe I8] que la creation volontaire de fissures A la ýurface r6duisait notabisnent sa duree de vie en fatigue, le revctement Ni-20Cr-1OAl-2Hf-O,LC, relativacent fragile, en so fissurant au bout d'un petit nambre de cycles, pout reduire consid~rablement la durde de vie de l'alliage 17], alors que lo revftesentNi-Cr, ductile, r.'aura pas cette influence n6faste.

7. CONCLUSION

Quelques aspects de linfluence des rev~tements protectours sur les proprietts mscaniques des pilces en superalliaqes ont dte discutos : notamment cycle toiriique de protection, interdiffusion entre te revetement et le superalliage, propri6tes physiques et m6caniques des rev~tements eux-m6mes.

Certains sCs offets decrits sont considdrablement amplifies dans to cas de la protection des pidces A patois minces fortement refroidies ; les problemos d'ir.terdiffusion rev6tement/superalliaqe, d'influence des revioýments Sur le camportement en fatigue thermique notan-cent, sont particulierement aigus et i1 n-existe pas actuellement do solution satisfaisante.

D'autreo- aspects auraient pu 6tre pris en compte dans cc texte ; ainsi la recherche de caract!- ristiqaes m6caniques Olevees pour les superalliaqes, soit pour augmenter la tempdraturo d'omp.loi des piices, soit pour augmenter leur potentiel. conduit S des coMpositions et des traitements theoriques d',,Iliages bion ddternin6s ; cpc choix prdcis peuvent conduire A limiter ceux do !a nature des revftements et do I. ,Cuthode de rdalisation, entrainant ainsi une tenue insuffisante de ces revdtcsents A lAoxodation et a La corrosion et par suite une !L-itation de td durdo de vie des pi~ces. Lour rdparation par renouvellement du revltement est couraute mais pose divers problmes : difficultds do d6capage pour les revoteoents 1 dhclliages, diminution de section des pidces ors du decapage des revdtemonts d'.luminiures correspondant A la partie de superalliage consarcd poor la r6alisation de ccs revftoments et limitant le nombre d'opFra- tions possibleŽs de renouvellement.

Les rdsultats pxsentes dans le cus des alliages IN .00 et iN 738LC no pouven, -aýo Ctre g Sno- ralisds S d'autres alliages ; chaque couple revotement/superalliage doit faire l'obct d' in-vest ;aLtuns particuliires dans des conditions aussi voicinoen que possible do celles c 1 'ut illsatlcT don-;ji ices.

i1 I R. 1,10CiR, O.M. 1180568 - RAPPORT 00857 N' 79-7-1419 (COST 50-217) Etude de l'iifloeoce des msthodes de protection et des revdtements sur It cosportemont des superalliases en fluage et en fatigue.

121 A. .TRANG, 5.1. COOPER - GEC l'EI'ORTN' -W/CmL/1980/4 (29.02,80) The effect of corrosion resistant coatings on the creep rupture properties of qaz turbine alloys.

7 [3I Dr. U• X!LIC. Pr. :CHNEIDER, l. VON' ARNIM - SBC-COST 50 (D2) ZWISCHYN6EV.EIC'IT 14' 3 t2 . 2. ASCHLUSSBERXCHT (20..2.81) N-ul ZB627-S]cN/NT/ý,rrs 108 Influence of coatin:3s on high cycle fatigue properties uf cast nickel base sujerallc!ys.

G40. LEIINET. N. SCI{IAIT - TEN-COST 50 (0201 F1NAl REPORT - OCTOUHRE 198 1 1ZBV57-Ci2t Ductility of metallic diffuoion typ, coati"gs on nickel based alloys.

SV Or. 0.1!. 5ONE - DOCUIMENT 81800 T;TMESCAL, JANUARY 1976 Overlay c,)atinjs for hiqh temrator avpj.I cai.ns.

jOI M. SELL, G.R. 0.11.iEVERANT, ;ELLZ" - ACIHIEV,ŽMENT OrF lilG6 FATIG;UE i-IU:TAIICE 11• -LETAL:S 810 *x.WLK'f. 8ATM s'rie 467, ERICAN StUCIL-Y FOR IESFING 85ND MATE.1A1.AL"2 ,ll , . I,1_i -153. l e-, . ots'ue•r.ntbof nickel-bas-e s•ueral loys.

17) D.I:. *RAMI IM. D.A. WO DFOuR - MET. TFANS, VCL. 12F, pl14tAoy 1'16 (__ -__,

The influencc of overldy c-atiiu-s - hi4l'ncyc:'le fatii . an advanced tantalun carl,itl, strenghthened eutou:t ic conp¶,sit e.

" " - l .1,• I,- 'il-f ;•ats;e --f asPi-'t 1i-i: Arqt..i-ao" "1 ms it -5) 0<,, mI Tableau I - Influence des cycles thermiques d'aluminisation IN 100 nu - ftuage h 85V7 C saus air.

ESSAIS COURTS ESSAIS MOYENNE DUREE

Cycle de traitt.'ment. c (MPa) ::t (h) (h) U(h)(M1,.-) :tt (h) .uptureurcpt5j e

1. Brut do caul" 372,8 116 200 294,3 515 927

2. 16h A 1050'C refroidissement 35C/mn 372,8 65 182 244,31 574 1129 (aluminisation basse i activit6 sur 6bauche) I 3. iWlow 2 + 30) 372,8 II_____74 222 ca-= A 850'C m cyc les de lh I j S= • 4 , idem 2 + 300 Scyc les de 1h ýk 100°C 372,8 1 71 151

5. 7,Sh 3 700'CCI .b 3,A OSO°C puls 9 treMpe 3 l'air 372,6 101 i 294 294,3 612 93% (aluminisation haute f activit6 sur 6bauche)

::temps moyens

Tablou Z[ - Influence des cycles therraiquesd'afuminisation INCO 738 L C nu - fluage a 850(C sous air.

Cycle de traitements ESSAIS COURTS

C (H4Pa) ::t (h) Il rupture

1. Brut de coul"e 294 9; 226

refroidissement 35"C/un 294 5 4 312

(aluminisation bass.. a.tivit- sur ,obauchd

3r fr idisseent 35%C par mn 294 5,)",

ll24h 3 845YC trempe a I'air (traitement thermique standard de 1alliage 3Ili)

4. idem 3, - 16 1t ) 10505C I refroidiss ement 35'C par .n 310 I 30 (aluminisation bass,e activitqý sur 4bauche) ii idem 3.

5. id,?. 3- + 1611 A 1053'C :refrlidisseýment 35"C par *,n 3 W l llB aluminisation basse activit:6 sur 6bauche)

l. ide, 3 + 7,5h 1 700'C

(aluminisation haute ac ivit,5 sur 31)31akich0 41 idem 3

temlpsroyens ill valeur irnterpolse Revett~meit ,NiAI

Phtses I:irigr4ruloves

o'•- 24,•, '.AlhageAIdq

Rev',e~rnie'- N. A.

Fig. 1 - interdiffusion rev~tement (aluminisanon basse activitw) sstratw a) 300 cycles de lb A 850' C sous Ar (image dlectranique) A) 300 cycles da lh A 105(1' C sous air (image dlect-onique) ci 30 cycles de lh A 1050' Cairus Ar (image A du chrome).

S.. . ..ii i ...... -...... ' .. i - 1 I I I 1 I I • 1'to I • | p

SRevetenme,*tl § ,NAI

S-'-X fi•-':•" =• PIorquettes U., ;)hrS'',

gi-

L3 J imNl00

Fig. 2 -- IN 1W0atuiwnii, (bane activite) fluage A 850 C soul air (388 MPa. niptror en 232 hi - phasesa o foana'es par interdiffusior,substrar/r'tev~tnt pendrant Ie main- lien S 850 C en fliage. .,~

.. . "_ a s e 1,31,,Ph cyle1hfug de d5-C.u ar(8 , rutr n/ 8

Rveem-, N,

Fig. 3 - IN 100apris aluminisatron basse activitJ, traitenent 300/hde 8 85C C saus arr t300 cycles de 1h), fluage 4 850 C sousatr (388 MPa, rupture en 984 h ).

" "Nei" -l Cle.r e'r.... N.

- .... • ,)C 2M~

i F;g, 4 IN$~~ousa 100yrgn spree C300 aluriia/n cyces d baSse lug seivt#ophi 850 traltern'enz C sect arr do (' 300hr388 8850 .4lPa, C I rmature en 2385h -- inraye elecrrorrsace),

Fig- IN lIC. ap , flurirrisaio5n bssou actrv(i 388M.rr uptuer300/if e 8 5hLuu r[0 8-9

-IZT7C 204'C 426C 646c 871'C C Allon gemenr a' 4l/u ,nrwure de "-r 3 . 6 n0rupturt (,) I ickol 1,75%AtJ

(Cc -22?2r - 12AI- 3Y)

2 iiD /ArD- 6 /'Co- FS,'r- 1041•-3)) jcLO-?.r-?OI-3,/'Alu'mninure / do cobalt 71. (5 A)

/ ~Fig. 6 - Eva lut~ion de la tempdrawure de transition ductile- ______Jfragile d'alliages Co Cr Al Y et d'aluminiuresN; Al r~et Co Al (dIapres [51).

o 200 400 0 80200 1400 19009600

4Loga (MPal

\ fupture 350 Allongome"*m ' \\ "%

t " . A11ag96bru de couliE - Ruplure • __ Trodlement 16h 6 I050"C 300 (refroldissernlW ?enf) Fig. 7 -Fluage .4 850- C sous air .log ui f (log 0 IN 700 brut . psultofs sur inrouvettes proligies ~- de coulde et traitd 16 h A 1050- C (cycle thermique de . . Risuiaots sur iprouveltes nuts cyant N \ \ 1rotec:ion). subs to Irctoment de profaction Log Alongement 1%. W 250...... -_ (hI FO 20 so 100 200 500 W0 2000

9."- • * ,,- * ~ J?- ' eveternlenk NAI

,, ft.

2Run)L1 ,uc refY'dU1&1V

30 t,m- IN 100

Fig. 8 -- IN 700 apris aluminisation basso activi:&, traitement do 300 Ih A 850ý C sous air (30X) cycles), fluage A 8500 C s~ousair (388 MPa, rupture en 98 h). -__-- -__-'

jt "

Fig. 9 - IN/GO non revttu. Fatigue a 850 C. 455 - 45-0 MPa, rupture en 240 000 cycles. a) surface de rupture (MEG) -_- - ...... c) cue en coupe cia l'amorce d. sup anorce, J 2 mn du plan m.d.an m~~ •. -" . S(,,coupe longitud..ale). 1. J' )

It A

Fi,4•Fg.10-I90 hoauiie aiu 5 ,4-450 lO~no rup~tureFatjI Is 50 cy4cl4eMs.

~a) surface do rupture (MESI - a

~~~~~camrc IMEvuB) op d 'aoc h) surface dMEu", de rupture(a diecismrc lcue (EG Inqtcise dirvtret

u d Ionituinae) -oe J.,

0 • '. ,,• A.v. -

• , -" : •I;i::"-

.•.r'• - It. .. ~ ., - ,, AA- . ,

C, 45 - 450 MPa, Rupture en 66 000 cyCIerS. Fig. 1 -1 IN 100 a7urninise haute activite. Fatigue J 850 a) surface de rupture (MER) b) surface de rupture do revitement (MEB) c) couxpe de I'amorce d) coupe loigitudinaledu revi'tement,

4MPa O,,,.

WN100. Fchtgue 6850*C, 5OHZ, (0111o0, Brut de coulie (usi'e) * Alumrtnse H A (NiAl) Chromalumrnise Aluminisi H A (Ni,•,A3 ) 500, •so: , . pt: .* . . .. .

350i "

300

.. 200

?00;

,So0 2 5 7 2C' 57 105s

Fig. 12 - IN 700, fatigue HCF J 850'C 9-1

RECONDITIONNEMENT DE PIECES FIXES DU TURBINE PAR BRASAGE DIFFUSION TURBINE STATOR PARTS REPAIR BY DIFFUSION BRAZING par/by Y. HONNORAT - J. LESGOURGUES D~partemcnt Materlaux et Procedes ODirection Technique SOCIETE NATIONALE D'ETUDE Fi DE CONSTRUCTION DE MOTEURS D'AVIATION 2, Boulevard Victor - 75724 PARIS CEDEX 15

Le brasage-diffusion permet, lorsque los surfaces a assembler sont tres proches (Icartement maximal de l'ordre de 504i. pour les superalliages), d'obtenir des zones de liaisons dont los proprietes sent du mnme ordre de grandeur que celles du materiau de base. Cettp technique trouve done une application de choix dans la rtparation des fissures fines sur los aubes de turbine apres fonctionnement. Cependant, comme certaines degradations depassent les ecartements maximaux admissibles, une nouvelle technique a 4KC deve- loppee. Ce procede, denonine "RBD" (Rechargement-Brasage-Diffusion), permet d'effectuer, 6 partir de poudres preallites, des liaisons sans limites geonrttriques. La zone reparee est d'une composition chimique lege- rement differente des pieces (a l'inverse du brasage-diffusion qui produit une liaison homogene) mais par ur choix judicieux des parametres operatoires du cycle de RBD, les propriotes locales peuvent 8tre du mWme ordre que celles dela piece massive. Diverses applications illustrant los possibilites en rfparation des deux techniques mentionn~es ci-dessus sont presentles. With very close surfaces (maximum gap of approximately 504nm for superalloys), bonding areas obtained by diffusion brazing have the same properties as the parent material. This technique is successfully applied for repair of fine cracks affecting turbine vanes after operation. However, a new technique has been developed to eliminate those defects which exceed maximum tolerated gaps. The new process, called "RBD" (Rechargement-Brasage-Diffusion), allows bonding without geometrical limits from pre-alloyed powders. While diffusion brazing produces homogenous bonding, the chemical composition of the repaired area after RBD processing Is slightly different from that of the part. But, rough part properties can be maintained in the bonding area, provided that operating paiameters of the RBD cycle are adequately selected. The two above mentioned techniques may have variuus repair applications which are presented.

PREAMBULE L'accroissement des performances des turboreacteurs est en partie assujetti aux progrEs de la metallur- gie et notamment 6 la determination de nouveaux alliages dont les limites d'utilisatior sont sans cesse repoussees, mals dont la mise en oeuvre implique le d6veloppement de techniques adapttes en particulier pour ce qul concerne 1assemblage.

C'est ainsi que lindustrie des turbomachines utilise largement des mattriaux tels que les superalliages de fonderie ou de m6tallurgie des poudres prdalltLes dont la "soudabilitA", par les proctdes conventionnels, est de plus en plus mediocre, certaines nuances comme le NK1SCATu (IN 100) ou le NCI4R8 (Reno 95) Itant mime reputees insoudables.

Aussi, des Otudes ont-elles 6t6 mentes par la SNECMA qua ont abouti a la mise au point de nouveaux pracides tels le Brasage-Dlffusion (BD) et plus recemment le Rechargmment-Brasage-Diffusion (RBD) qua, outre 1'avantage d'itre insensibles aux problmes de fitsuration, prtsentent un intorft Oconomique certain de par laur rOalisatlon au four sous vide industriel, sais outillages spkcifiques.

CGs techniques trouvent un terrain de choix dans le domaine de la reparation des pieces de turbine oD la nature, la morphologie et la qeantite des degradations en fonctionnement sont tres variees. Nous nous proposons, apres une presentation des proc6des BD et RBD, d'illustrer par des exemples traltts & la SNECMA les possibilites offertes par ces deux techniques.

1. PRESENTATION DU BRASAGE-DIFFUSION (BD).ET DU RECHARGEMENT-BRASAGE-DIFFUSION (RBD)

Le Brasage-Diffuslon a Ato defini comme un procede u'assemblage par un metal d'apport qui, par le jeu d'une phase liquide transitoire et de la diffusion inte-mýtallique, permettrait d'obtenir des liaisons homogines chimiquement dont les caracteristiques sont du mWmeordre que celles des materiaux a assembler. Cette possibilit6 suppose un choix trýs precis des parametres optratoires qua sont principalement :

- la nature et la quantito du m4tal d'apport qua, outre son rOle de remplissage du joint, doit etre compatible chimiquement avec le metal de base )our permettre une diffusion corplfte et r~ciproque des elements en presence, - le cycle thermique, moteur de la diffusion, qui doit Mtre convu de favun A Aviter les formations de composes stables, les retassures ou las digagements gazeux et permet~re une houmotn'ilsation aussi complete que necessaire de la zone de liaison. Sur le plan technologique, le metal d'apport peut 9tre depose sous forme de poudre ou introduit &jns le joint sous forme d'ui Ieuililard. Lu fijurtI' a .;;crC.cher.atiquement iif"rcris pwrts et retracc los differentes etapes du processus. S ~9--2

Fig., IPRINCIPE DES PROCEDES BRASAGE-DIFFUSION (OD) ET RECHARGEMENT 1BRASAGE- DIFFUSION (ROD)

Fig. 1a: Brasage Diffusion Fig. 1b: Rechargement Brasage Diffusion

Application du metal d'apport Melanige do poudres superalhage base Ni ou Co Posidr. do Poude deet ,51a; diipport Feudlaid me!.1 d apport

Je, opa's Jet fi

Fusion du m6tal d'.pport

L .. Solidification isotherme -

Diffusion (homog6nibisation chimique)

~ r;r

Pour les superalliages base nickel et cobalt, il a pu ftre rnontre que o'Lxcel1ents rcsultats (.taient obtenus en utilisant certains metaux d'apport fondes sur les systmnes Ni-Cr, Ni-Co, avec des additions de B eL/uu 'Je Si, dort 1a ,utrice est d'unu coe-•position voisine de rplls dps materiaux a assembler, et vourvu que le jou entre les surfaces a assembler n'exc¢de pas une valeur de 50 a 100jim selon les alliages.

La figure 2 illustre un cas d'application obtenu sur un alliage NKI5CATu (IN 100). Sur cet exemple, le uAsage-diffusion a ete realisz au four sous vide a lalde d'un mftal d'apport du systeme Ni-Co-Si-B ; le'tude sntallographique revele que la diffusion est totale, resultat confirme par lessai de fluage rupture ,1 980%C qui montre ime te;iue .?s i •c:s Ju MC-, o dlu quC .&i le du, m.tri.3u de.base. 9-3

Fig. 2 BRASAGE-DIFFUSION DU NKISCATu (INI100) A LAIDE D'UN METAL DAPPORT NiCoSiBT_

0 100'4"

*. .

:: S,

-a# I

2. EmI. do fluage rupturo 6 9800C do co•W6e 0 en MPa .... M6 do bse *t it 2- -- -" Fprouvettes assem~s en bout 6 bout 200

100.- - - Temps de S.rupture 10 -0 20 40 100 200 1000

D'aussi bons resultats sont obtenus pour la majorite des superalliages de fonderie. Ce par se, ca3rct.ristiques, le brasage-Diffusion (8D) est donc bien adatý a la r~ppration des fissures fines qui proliferent souvent en rýseaux denses : les ph~nomenes de mouillabilite et de capillaritU permettent d'assurer un bouchage physique des dfauts puis le travail m~taIlurgique de la diffusion restaure en ces zones les proprietts du metal de base.

Cependant, la limitation dimensionnelle imposee dans le jeu des surfaces A assembler ne permet ni d'eiivisa'jer ]a r~p aration de toutes les degradations possibles, notamment sur une aube de turbine, ni de recharger les SUrdL, , .rodees. Aussi, une technique a-t-elle ett d~velopp~e par la SN[CMA a partir du brasage-diffus-un (BO), le "Rechargement-Brasage-Diffusion" ci-apres denomni RB0.

- 9)4

Darts son principe, le RBD (figure 1.b) consiste a densifier au cours d'un cycle thermique, un mW1ange de poudres prealli~es de deux categories distinctes : - l'une de conposition analogue a celle d'un superalliage base nickel ou cobalt classique, - I 'autre de composition analogue A celles des matiriavx de brasage-ffusion, la composition chimique globale de l'ensemule etant identique ou tres proche de celle ou materiau coistitut" f des pieces traitees.

La diffusinn des elements abaissant le point de fusion des poudres de la deuxi•mm categorie se fait, au cours du cyCle de traitement thermique, autant vers les surfaces de la piece traitee que dans les grains de poudre de )a premiere categorie. Cet artifice permet, par rappo,-t au 60, de raccourcir lts parcours de diffusion et de conserver des temps de traitement raisonnables - e. identiques - auel qun soit le volume de matiere diffuser.

Fig 3 ASSEMBLAGE PAR RBD DU NK15CADT (Ren6 77) A LAIDE DUN ALLIAGE RBD[NiCrB * NK17CDAT (Astroloy)1

1. Aspect micrographique do Ia liaison

4- -=4;--=

•== rR

•=am{

2. Essais do fluage rupture 6 980*C

- __~ Metai de base that TR

(Y anWe Eprous'ete assemblee en bout a hot

140-___ ------

120,- -

100 - - --

Twnps d. S0. - .. wore en h 10 20 40 100 200 - 400 1000 2L'exeDle de la figure 3 montre un assemblage realise sur NK15CADT (Rene 77) a Vaide 'un ,at.riau RBD elabore avec un metal d'apport NiCrB et d'une poudre •;KlICDAT (Astroloy) conduisant a wre coiDootition finale Sprohe de ielle du Rent, 77. Seule une variation structurale due au bore exceoentaire peuL y 4-te remarqutt. *- difference qui nentraine pas d'abattement en fluage-rupture A L.essai Am a at•ie tie.-•oU4 lf~iore -1 d'eprouveLtes prismatiques reconstituees par RED ne fait pas apparaltre d'accroissenenýt de vitesse de fissuration dans les zones de liaison.

11 est possible d'obtenir des structures derses pratiquemznt sans limitations q,onetriques, s(uit dans do, interstices entre des pieces a assembler, soit sur la surface externe d'une piece. De telies structures -ssurent une continuite, dans tout sor volute, des caracteristiques mecaniques et physiques de la piece trai tee.

F,_ 4 ESSAI DE FATIGUE THERMIQUE SUR EPROUVETTES EN NK15CADT (Reni 77) RECONSTITUEES PAR RBD A LAIDE D'UN ALLIAGE RBD[NiCrBO NK17 COAT (Astroloy)f .

1. DWfinitions

a) Eptouvetles

~ieiapportore ps R8D

soc do b) Disposdif chalumeau

Eprouvetts Moo.vemnem

C) Cycle

Chauffseg: 60 sec 9 1100C Refroidissamert: 20 sc Bu~e dir cofnilwit"

2. Rfamitmts dos *"ais .• Eprou-tte.massive (jrfiremebase) S- • •Fissure NWi deo i'ptotw*ete assemblhu

Longu.ur des fissures on mm Fissure W2 do l~prettesseribe

do cycles 0 100 200 300 400 Grace au Jeu des paramftres ojtratoires qui i•issent les prnpriCtks physiques du melange pendant 5 'o$- ration de c.nstitutlon, il est possible de bitir des furies, cortb',er des trous, recharger des swrfaces...

il y a cependant lieu de roter que lorsque le volume concerre par le matý,riau RED devient trop inportH:, r, c'est-a-dire au-dalk de qtnlques dixiAmes de millimýtres, l'influence du matkria-, support s'estormpe et 1a zone R8D pr~sente alors sa structure propre qui est cell d'ur, ;rwttriau "mAta)lurgie cdrspoudres', ;,reirs "equia-es moyens & fins. De plus, grace a leurs principes identiques les prccedes BD et RBD peuvent 6tre combines, cest-A-dire que les cycles thermiques propres a chacun sont compatibles, peuvent se cumuler, se combiner..,

11 apparalt alors possible, par une d~ter-,ination precise et une combinaison aud'cieuse des op~raticrS, d'aborder tous les types de r~paration sur pieces de turbine, comme en timoignent quelques exerpias traitiýs A lA SNEZ'%A et presentes ci-apres.

2. MODELES D'APPLICATION DES PROCEDES BD ET RED EN FONCTiON UL LA REPARATION EN)VISAGEF Les degradations essentielles qui caract.risent une piece de turbine deterioree en fonctionnement sort - d'ordre geomtrique (fissure, arrachemert par irngact, erosion),

- d'ordre structural (corrosion, brOlure). Devant des d~gradations du premier type on procede yýn~ralement a un reconditionnement des surfaces de fagon a assurer leur mouillage et a une liaison mdtallurgique. sans enlývement de matiCre. Dans le second cas, on procede & une flimination des parties dtfectueuses et d leur remplacement. Cette technique de "rapievage" s'applique aussi de fagon avantageuse lorsque les dommages de type geometrique sont trop etencus Ou quo la forae de la pidce a reparer est trop complexe. Les exemples suivants sont destines a illustrer les possibilites offertes.

2.1 Reparation directe de la piece

Ce type de reprises est pratiquement limitd a la reparation de fissures, au rechargement de petits manques de mati~re, de parties drod~es. Un premier exemple est celui d'aubes fixes de turbine, coulees de precision en alliage a base de cobalt KC25NW (HS 31), dtgraddes par fissuration ou fatigue themique, Pour le reconditionnement de ces pieces, il a dti necessaire de procdder & un rechargement du bord de fuite en vue du ragrdment du profil,

La rdparation a 6t0 conduite salon !a gamme de principe suivante :

- nettoyage des surfaces par une succession d'optrations classiques telles que d~capage chimique et parachevement par un traitement sous atmosphere contrOlee, - remplissage des fissures les plus fines par un metal d'apport A base de NiCrB sous forme d'une pate constitute de poudre prdalliLe et d'un liant volatile, - fusion du metal d'apport et colmatage des fissures par passage a 1200% UM5 sous vide, - depOt !'un mat~riau de RED constituA de poudre d'alliage base cobalt et de poudre de m6tal d'apport du type NiCoSiK, - fusion du metal d'apport par passage & 1200°C 15 mn sous vide, - contrOle de laspect g~om~trique des r~parations, - traitement de diffusion a 1200%C 4 h sous vide.

L'aspect d'une de ces pieces est montr@ sur la figure 5, a l'Atat brut de reparation. La microphotogra- phie met en evidence, pour les zones rechargies, la continuitA de la piece a la reparation et la nature de celle-ci.

Un autre exemple est celul d'aubes de distributeur de turbine, en alliage a base de nickel NC22DK (C 242) pr~sent6 par la figure 6. Dans ce cas, des fissures assez importantes Otaient prhsentes sur 1 'intrados et la reparation a Literealisde par one operation simultande SD/RED A laquelle un rechargoment RBD a etd adjoint Iour la retouche du bord de fuite. Du point de vue r~tallurgie, nn a obtenu

- une parfaite continuite entre la rtparation et la piece, - une legere difference entre les structures, le metal de base, un alliage de fonderie base nickel, prisentant un grain plus fin que le rechargement.

2.2 Changement de parties defectueuses

I] arrive, dans certains cas, que des degradations trop irportantes conduisent a opdrer un changement total des parties defectueuses. Cette solution peut egAlement Otre prefdree lorsque les zones concernýes sont trop ouvragees pour subir des retouches directes, comme par exemple un bord d'attaque perforQ d'aube mobile.

On procede donc au decoupage de la zone endomnag~e pour y substituer une partie neuve de mwme g~om, trie. F g S REPARATION D'UNE AUBE FIXE DE TURUINE EN ALLtAGE DE FONDERIE KC2SNW (HS31) PAR OD ET ROD

<:j

-tt

d,• bor~d de|t tki~te

R, KC.',NW .

I -* , REPARATION D'UNE AUBE DE DISTRIBUTEUR DE TURBINE EN SUPERALLIAGE BASE NICKEL DE FONDERIE PAR 50 ET RBD

m I alp,-

JIM

Aspect du RBD stir le bord de fuite La reparation peit ýtre abordee selon deux voies distinctes, en forction .e i'objrA!! A :

I !a reenerine ces cairdctristicques 'k-afli()cS eý qe 5 c n½ Cf..c i-al I-:,' parfaite continuit$ de la structure, necessite l.e reours au oras c-ff,.sic, c- " - aussi parfait que possible de• surfaces I assret'r ; ct c"- srea r erc~e ar;-ft-re pour le Cnangement u'urn bori2 d'attaque ou Ac fuite sar 6 aute mobile de 'trbine oa la cualite deS liaisons doit Ctre aussi bonne que possible.

Cuonte enu de la geoetrie des pieces, on lrocede generalement a ur usirage mrcanique des surfaces d'accostage de fagon A assurer des tolerances etroites.

2 La reconstitution geomntrique au mnoindre coOt. On peut alors s'accomoder genera -ee•td'une legere evolution locale des proprietes mecaniques et done d'un accostage moins precis des surfaces a assembler. Le choix se porte alors sur une ablatioi, des parties endommagees par usinage EDM au fil et un assemblage de l'insert par RbD. Cette deuxicme voie est ,liustree par l'exemnple donne figure 7 dU changement des DordA_d'aetaque et de fuite sur une aube fixe de turbine en alliage A base de cobalt KC24NWTa (Mar M'539).

Pour cette application, un montane peonet ce )nettre en position les pisces, en vrue d'un ! intier. par pointaqe ; le jeu releve qui varie alors de 0,2 a 3,: nu est cotble par 1 'alliage ObD co,•sti- tue de poudre e NK1ICDAT (Astroloy) et di'n metal d'apport "iCoSiB. Dar.s ce cas, i! a rte juge preferable d'eviter l'utilisation de poucres d'alliage a base de cotalt susýCepticies Ae fcr-er, ions du traite.ent thermique, des composes intermetailiques se co-portant co.n-e des 2arrieres Ce diffusion. Le traitenent de diffusion est effectue en four sous vice sans outiliaýge partic; 'icr.

Sur des secteurs multipales, il -st possible que la degradation .E l'ure des pales soit telle que son changement total devienne oiligatoire.

La recuperation peut alors 6tre envisagee selon le principe sulvart : les secteur' ref-m' sont decoupes en aubes elementaires, par les pltes-fornes, a ces cetes xtarcaris, par ece Ciesur An- machine L.D.M!. Aaidea d'un fil de I run d'epaisseur ; les pieces ur.itaires sont en7s-te sciL rebut~es, soit recupdrees pour les elesents suffisa~nnt sains. ! oartir de ces cernnerr ce nouveaux secteurs sont construits par assemb~age RZ-.. La preparatirta 1 assemolane consists esseC- tiellement a pointer les piices sur on nontage qui assure 1a sec•tian Oe pass?-e lal veilne. -,'es reamplissage des joints par un materiati RED, i'assemblage lui-kimc cat 'rnen au four sous vide. type d'operation peut 4tre effectue en atelier apr~s ý-.integration des pieces refornes. C.ne rt-oa- ratiOn identiqUe pout #tre effectuee sur des Polces neuves reoutees en fonderie pour defauis ccc- reetnricues. Un exemple de cette possibilite est montre par Ia figure .< qui il1ustre la recorstitution .*esecteurs de distributeurs de ttroine en alliage NK'iCADT (Rene 7,. Los pnotnorapnies •ontrent J.,;ur entre ies surfaces d assembler apr~s le pointaqe, et l'aspect externe Ae ia liaison RBK cis• suffi- sannent fin pour eviter toute retouche mecanique.

Conjointement a Vassemblage, dans le cas des pieces reparees, certaimes porti~es scnt recnargees afin de pernrettre on usinage prkcis avant remontage. "Dans le cas de degradations i-portantes, il est aussi possiol,' de reconstituer enticrenent la partie degradee a partir de poudres et lexemple prosente a la figure 9 illustre cette possibilitY. II s'agit ici de la reconstitution d'un bossage de prise d'accessoire sur une aube fixe de turbine eft NKISCADT (RenA 77). Dans ce cas, la proeminence initiale necessiLant one modification geor-strique a ete arasie et une nouvelle forme a ete btie, directement en materiau R5D. Four ce faire, tnn forme en poudre RBD a Lt@ prefrittee A une geomtrie tenant compte Au retrait ulterieur du materiau ions du cycle thermique . La forte de la prise d'accessoire est ensuite obtenue par usinage. L'inter~t d'une telle operation, reside essentiellerrant dans son caructbre econo.-ique dO ý la faci- lits d'application ; la retouche est en effet appliquee directe-,ent sur la peau de fonderie apres tine preparation chimique, voine uin simple degraissace si ia surface W'est :ss oxey~ee. De 'tori.re:.ses interventions de ce type sont possibles : bouchage de trous, reprisas de nervures... poarv- que les caractaristiques recnerchdes aient pu Otre 'uparavant assurees par des essais en Laboratoire.

COS'2LUSION

Le developpeuent par la S'ECMA des techniques de Brasage-Diffusior: et de ecnargenet-Srasanc-Diffuaior pernet une nouvelle approche des problemes de reparation. Leo pre;oie'es applications ort rte motivees par la carence des procedes de soudage conventionrel notasTent sun les surerailiaccs Ac fonderie. Mais l'int~rnt economique est site apparu e la reparation de pieces s'autres type•s C<..nb,? de combustion, carters...D. Ces techniques etant mise en oeuvre A 1'aide d'un fc:ur sous vide Je tras- tement industriel et d'outillages simples, elles perniettent le traitecect simultoane Ac nor,-reuses pieces.

Bien entendo, une experience reste a acquerir qui perrvittra c'affiner Tes techmniques et d'en obtenir un meilleur profit. Nous pensons par cos quelques lignes, avoir cis en evidence l'etenoue ies possibilites offertes quo-que la liste dt 4aý traites spit loin d'ýtro -ihaJstive. r,•7 CHANCEMENT DES BORDS DATTAQUE ET DE FUITE SUR UNE Aa, AUBE FIXE DE BANC EN KC24NWTa (Mar M509)"CO

us - -

144

:, • ,-i 2i;,r'.... .'. Fig 8 RECONSTITUTON D'UN SECTtUR VE TURRINE EN SUPERALLIAGE DE FONDERIE PAR ASSEMBLAGE OD ET ROD DE TROIS AUBES ELEMENTAIRES

Pieces point6es on position avant assemblage

Aslpect upries ass.ainlaae du sec:taur reconstitu6

", "at. IIr -Fig 9 IMPLANTATION D°UN BOSSAGE DE PRISE DACCESSOIRE SUR UNE AUBE FIXE DE TURBINE EN SUPERALLIAGE BASE NICKEL NKISCADT (Ren& 77)

[]~' "'Wil RI;Mi:

i i =

Aspor'{: (iii pfvfat

i| I. - A!c lhi|,q n o~e

* * x i i i I' . ...-i REJUVENATION OF USED TURBINE BLADES BY HOT IOSTATIC MPESSIN' AND REHEAT TREATMENT

CHEUNG, K. L., LEACH, C. C., WILLETT, K. P. WESTINGHOUSE CANADA, INC. P.O. BOX 910 HAMILTON, ONTARIO, CANADA LAN 3K2

KOUL, A. K. NATIONAL AERONAUTICAL ESTABLISIIBIENT NATIONAL RESEARCH COUNCIL OTTAWA, ONTARIO, CANADA KIA OR6

SUMMARY

Gas Turbine blades operating at high temperature and stre,.s experience microstructural transformations which eventually lead to, their replacement. The occurreuct, of creep beyond allowable limits is a maS"r cause of turbine blade replacement.

Microstructure and properties oa Many superalloys can be resotred by reheat treatment. This restoration Is only partially effective if internal creep voids ire present. Hot Isostatic Pressing (1I1t) Is a process which can be used to leal sumh creep voids and, in conjunction with the reheat treatIment, provides a viable metlhod otr the rejuvenation of used turbine blades.

INTRODUCTION

The use of rejuvenation processes, usually beat treatments, to recover the properties of superalloy materials is a concept that has been with us for a namber of years. However, this approach has received a new impetus, particularly for gas turbine blade materials, due to the current high cost and decreased availability of these components.

A major cause of blade replacement at overhaul is blade growth beyond the acceptable limits. Excessive blade growth, or creep, may be due to a number of usecl,aismsr. Blades which experience a turbine overtemperature can undergo microstructural transformations which reduce thu creep resistance of the material. Such blades can be expected to grow at a faster rate than those with normal mlcrostructures. Similarly, blades with a long in-service life will experience this lessening of creep resistance, albeit at a more gradual rate. The normal blade tip clearances are quite small, and often as little as I or 2% creep elongation in the blades can cause tip rubs.

The degree of microstructural degradation found at overhaul may be acceptable, and in such instances the use of repair welding, blending, etc. can be used to return these blades to service. However, the blades will have lost some fraction of their useful life, and would benefit from a rejuvenation treatment. Further, if creep voiding has occurred, heat treatment alone will not heal these cavities. Here the additional factor of pressure, found in the hot isostatic pressing (HIP) process, is required.

There are several benefits from a successful rejuvenation process. The ability to return scrapped parts to service has obvious economic advantages. Reclamation of used blades decruases the need for spare parts, a definite gain in an era when suppliers must meet an increased demand for new components, to which the spare parts inventory perforce takes second place. Also, a reductio:. in the need for superalloy spare parts assists in the conservation of relatively scarce materials, e.g. cobalt, tantalum, etc.

The present progranose is aimed at developing HIP rejuvenation re-heat treatment cycles for two Nimonic alloys, Nironic 105 and Nimonlc 115. The development of the cycle' is in conjunction with the National Aeronautical Establishment, National Research Council, Canada.

REJUVENATION CRITERIA

Turbine blade life is limited by internal and external damage. The external damage includes hot corrosion, ox!dation/eroslon, thermal fatigue cracking and foreign object damage. The environmental effects are influenced strongly by the type of fuel used: land-based turbines utilizing clean natural gas suffer littie or no corrosion attack, while similar units burning oil or similar fuels will experience corrosion early In their life, and must be monitored more frequently. Litlin limits, oxidation and corrosion products can be removed successfully by blending, hut the repair of open cracks on the blade surface is not so reliable.

The internal damage which occurs in the nickel-based superailivy turbine bladts many be brcadly divided into two types. Firstly, microstructural damage results from the alteration of the basic structure due to high temperature thermal cycles. The nickel-based superalloys depend upon a fine dispersion of the gassa prime precipitate within the gamma solid solution matrix for their elevated temperature creep resistance. Thermal cycling results in the coarsening or agglomeration of the gama prime phase, resulting in fewer obstacles to dislocation movement, with a corresponding reduction of creep resistance. Other undesirahle effects include the breakdown of primary carbides and the formation of the topologically close-packed pbases, namely sieta, Laves and mu. These microstructural changes, either can be reversed or have their effect minimized for many alloys by re-heat treatment, or ' lermal 'ejuvenation cvcles.

-- l-A S~10-2 0 The other area of internal blade damage may be classified as structural discontinuities. These

Include cavitation, micro-cracks, and, In cast blades, Inherent casttng detects. caviratlon is the term applied to internal microporosity at grain boundaries with a transverse orientation to the direction of applied stress. These cavities can grow and i,0- tip to •orm micro-cracks, and give rise to arstress rupture liiltiro. Micro-cracks may torm also at such inrernal stress raisers as inclusions, er from Lhe fracture of brittle carbides. Casting defects, of courset do not result from turhint ,peration, but are tormed dur!ngsofildticaaion. Porosity, either dueac to .hrinkage,Lraippi d as or is the Most C smnonCasting detect, bupocs wuch detects Cao havei a deteriou slitect on LoheoPf m ating liwer of the blade.

ITh- aoncuggeptd w arlierd t t ine majority of icrostruclural defects were the resultm p elevated temperdture, anld may be reverted ',v licit treatment. Similarly, structural dizicontinuitiesi, other than Casting dilect.se aro gentrally thes rtralt et applied-vtrersef. These detects cno we seanad by the uae ot stress, the pressure tn to HIP cycle. Therefores a combination and heat treatmest:itn1P Can be Deffectivein reversing both elarsificacp sed of internal0lade dadage, and restore not only the microstructureiut also hathe hanical properties of turbine blade r eateriols. SltIP FACILITY

ther ot Isnistatlc Pressing process was developed originally ior theo consolidationp metal powders to attain theoretical den simplicIt uit iP reucally a pressure vessely Figure ltwhich is pressurint ed with ale inert gas ac d simultaerusey heated. The corubinasion ir reat ad pfeaures cotlapsoer internal caviting t rjdbonds surlaces tog, ther.

eif!s concept was applied to investment cast turbine bladei to heal the internal micro-porosity inherent in these coraponents. Tefs treatment proved very effectivhe and it is now standard practice at Westinghouse Canada Inc. tallowP as-cast ulades iu IN7o38 material before installation insurbines(t). REJUVENATION PHlILOSOPHIY

Data accumulated on service exposed X-7sqtncooel turbine blades with microscopic voids show that full resioration of mechanical properties iý possible with a eP re-Incatd treatment rejovenation Figure 2, Frther, in-service myestig of time treated bladesFhows no degradation of properties after 2sv000 hours of post-fejuvenation l tfeFiguren 3p(3),

mespite the relative simplicity o the iP rejuvenation concept, early work(3)hasdur own that careful selection of cycle partmeters is essential if the requisite mrcrostructural features are to be maintained during tae rejuvenation processd as e need for the optimization of oIP and/or post-oIP thermal cycles for specific applicatpons fannorbe over-emphasized.

HIP CgCLE

The HIP temperature selected for a given alloy is above the gamta Frimi solvus oemperatureaor the alloy. Such a temperature allows for teie dissolution of the gamma print phase, and ensures that material Sflows readily under tihe isostatic pressure imposed to seal the internal cavities. Other factors must be tconsidered also for the exact temperature o be chosen(. These include the grain coarsening temperature of the alloyr which may be time dependent also, Figurei. Another considerdaion is the solvus temperatures of precipitates other thand gsI a prime. The ideal selection would include examination of virgin alloy tmateria to determine the degree of primary carbide degeneration that has occurred during service exposure. Furthero much rejuvenation work Is naturally carried out on older materials, where perhaps the composition was not as strictly controlled as in later alloys. The examination of virgin stock could determine the presence of sigma phase, for example,and how much, if cay, was toleratedin the original blades.

HE-HEAT TREAT"MENT CYCEI

The aim of the post-HIP reheat treatment cycle is to restore the mpcrostructure of the alloy to its pre-serv ee condition. Som e a slow furnacntot foone ce f cbe solution treatment temperature to either a lower ageig temperature, or a partial solution treatment temperature. Such a treatment exists for Nimonicfra'(ti ), and a similar treatment is used for siPprocessed IN-738 castings(e). The treatments give improved creep ductility over the conventional air cooling trolledmts. Thin is attributed to t ueformation of serrated grain boundaries, that are believed to resist sliding and give rise to more homogereous derforuatnion.

The slow cooling rate can provide rple yrphogy,grain and twin boundary and by allowing longer times for MW:antd M23C6 particle growth, could cause tthe formation of large, discrete carbides at these boundaries. Such larger carbides could improve creep rupture properties by disrupting grain boundary sliding.

"The adverse effects of the slow cooling can be the formation of carbide platelets at the grain boundaries, and a Larger gamma prime particle size, with a .nterparticleresultant increased spacing for a given volume traection of izamna prime precipitates. Such particle size changes tend to increase thle cr e r ierotrcur as atresult of easier Oroean loaping and difusion controlled graine boundary migration(ttr

Thu thr s•rlcction Lit the po:;t-,!IP reheat treatmenIt cycle mu. t c• dr-

a) !,created grain boundary formation b) carbide precipitate morphology c) size and shape of garmma prime particles.

The microstructural features are interrelated, and the c-ptimum results are obtained by C,,ntrol •L1 "solutionl temperature, cooling) raLv, partCial Ikelutioln tem~perature and ageing treatment. 'a 10-3

Cum"R DEVELOWNs

FHIP reheat treatment rejuvenation cycles are being developed for Nimonic 105 and Nimonic 115 Service exposed turbine blades. The blades had experienced approximately 100,000 hours of exposure, and 4 the as-received materials showed a corresponding degradation of physical properties. This was particularly noticeable In the Nimonic Il% blades, where hýeavy sig=m phase for•ation had cauaed severe embrittlement.

Initial rejuvenation trials have resulted in the restoration of tensile and stress rupture properties but full ductility has not been recovered. However, further- cycles involving slow cooling tec;hniques are being studied, and a full restoration of all properties is expected.

The second stage of this programme will be the rejuvenation of a complete row of service exposed blades of each alley. The treated blades will then be returned to their respective turbines, and their in-service performance measured on a regular basis. This monitoring is essential to prove the long term effectiveness of the rejuvenation treatrtwnts.

CONCLUSIONS

The microstructural philosophy necessary for the successful HIP reheat treatment rejuvenation of service exposed superalloys has been developed, and Westinghouse Canada Inc. has applied this philosophy to used turbine blades.

The in-service performance of rk "uvenated parts has been monitored, with very good correlation of properties between rejuvenated and new oomponents. Existing results suggest that HIP reheat treated blades are a %iablealternative to new reF/acement blades for gas turbines.

REFERENCES

1. VanDrunen, et. al. "Hot Isostatic Processing of IN738 Blades" AGARD Conference, Italy, Sept., 1978.

2. VanfDrunen, G. and Liburdi, J. "Rejuvenation of Used Turbine Blades by Hot Isostatic Pressing" 6th. Turbomachinery Symposium, Texas, Dec., 1977,

3. Liburdi, J. "Rejuvenation of Used Turbine Blades by Hot Isostatic Pressing" ASME Int. Gas Turbine Conference, Calif., March, 1979.

4. Koul, A.K., and Wallace, W. "Rejuvenation of Turbine Component-Microstructural Considerations" LTR-ST-1235, National Research Council of Canada, Feb., 1981.

5. Betteridgep '4. and Heslop, J. "The Nimonic Alloys" Edward Arnold (pub.) London, 1974.

6. Hagel, 14.0. and Sims, C.T. "Superalloys" Vol. 1, Chapter 4. John Wiley and Sons, Inc. (pub.) New York, 1972.

7. Beddoes, J. and Wallace, W. IMtallography, 13, pp. 185-194, 1980,

8. Bilsby. C.F. Ph. D. Thesis, Cambridge University, 1966. 10-4

. .I.I SENT T,,E.OEO TOP CLOSURE

OiI~nG SE AL,

... .VE:SSE L C OOLlINdG JAC KFT '[p•cimEN TrIC(6)-- -

LOAD FIXTURE IN 430 -' CIA. INSULATE D HOOD "; I(•0 m LONGO SWORICZONE ! "" PR•SýIJRE" VE:SSFI.

GrIAPHII I fPC TAL -

GRAPHITE tILLME NTS (51- -

.• __- CERAMIC BlASE:

B-07TOM CLOSURE

-ORING SEALS

-RESILIENS' THREADED BOTTOM NUT ELECTRICAL LEADS

VACUUM ILINE

TIC LEADS

HIGH PRESSUREI.INE

FIG.A. SCHEMATIC OF THE WESTINGHOUSE CANADA HIP VESSEL

.IA. K-,/ 51"

S. TI> '1' 41'f pI"

FIG.2. STRESS-RUPTURE PROPFRTIES FRIOM SrVERAL SETrs OF INCONEL ALLOY X-750 BLADFS AFTER VArIOUS TREATMENTS COMPARED TO AVERAGE NEW BLADE LIFE MCO'C. &LLOYX-?2O TCRA"iOI lOAT

~o400- FROMr M.CML tl I)

r.4- I-LAI\

2 00-

I_-I

20 21 22 23 24 25 PARAMETER P

FIG.3. LARSON-MILLER PLOY (P = TX 103 (20 + log t), T IN 'K. "tIN HOURS) COMPARING STRESS-RUPTURE PROPERTIES OF SERVICE-EXPOSED, VIRGIN AND HIP PROCESSED BLADES

S==--•250 - 250 AGED 870'C/124 h1 *l AGED 870'C/,00hr

200-

-. 15o-

L.J -j

M 100

50-

0.2 0.4 OG GRAIN SIZE mm

pG,4. MICROSTRUCTURAL DEPENDENCE OF RUPTURE LIFE OF SERVICE SIMULATED INCONEL 700 TESTED AT 79()*C AT 345 MN/mr

mEL o sr.. 110C A S.T 1205'C 0 AS RECVIVEO ST 1180'C

SI0-0- 1-7A

.li. ii

!iu !L)

000m 0.04 0,08 0.12 0.16 020 0.24 aiiiQ xY'PARTICLE SIZE pr~n

iiA

•= FIG. 5. -f' PARTICLE SIZE OEPENDENCE OF CREEP RATE OF SERVICE ii -a______i______SIMULATEO JNICONEIL 700 ,lIe 00-0 08 Y ATILiIZ012 x 01 00 02

FI..iPiIL IEDPNENEO RE AEO EVC HIP PROCESSING - Potentials and Applications

by W. J. van der Vet CHROMALLOY DIVISION-OKLAHOMA 1720 National Boulevard Kidwest City, Oklahoma 73140 U.S.A.

SUMMARY Introduced in the early 1960s as a sophisticated metalworking and bneat treat process, hot isostatic pressing has made strong inroads against estaLlished techniques. Hot isostatic pressing, or HIP, is a process in which components are subjected to the simultaneous application of heat and pressure in an inert gas medium. From its inception as a means of gas pressure bonding nuclear fuel assemblies (1) HIP processing today finds a wide variety of uses from cermet cutting tool manufacture, to the production and rejuvenation of gas turbine hardware. (2) The results obtained on complete engine sets of Inconel X-750, Udimet 500 and Rene' 100 turbine blades indicate that HIP processing is capable of restoring new or near new creep properties and low cycle fatigue properties to tsed blades. The process should be applicable to most superallcys, nevertheless, recent work emphasized the need for preproduction process parameter verification to establish optimum cycles which take into account both the metallurgical and mechanical aspects of suprralloys.

1. DESCRIPTION OF CHROMALLOY'S HIP UNIT The basic equipment needed for HIP processing concists of a pressure vessel, a high temperature furnace, a gas compressor, and a temperature pressure control system. Ad- vanced HIP pressure vessels must withstand pressures up to 30,000 psi (2068 bars), furnaces must operate up to 2730 0 F(1500°C), the gas comp:essor must not contaminate the work ' ng flu~d, and the control system must maintain work zone temperature to within + 25 F(+ 14 C), and pressure within 2%. Figure 1 illustrates the latest generation of equipment, and Figure 2 shows the actual Chromalloy HIP unit. In Table 1 the basic parameters for the Chromalloy HIP unit are given.

The Chromall 8 y HIPofacility wll accgp': parts 18"D x 20"L for processing at temperatures from 1400 F(760 C) to 2400 F(1315 C) at pressures from 5,000 to 29,000 psi.

The unit offers complete thermal cycle capability including rapid quenching. As will be discussed later, this feature is particularly useful for engine run turbine hardware which requires post-HIP heat treatments for optimum property recovery. Use of high purity liquid argon as a medium for pressurization minimizes the potential for surface contam- ination. Temperature of the work pieces is precisely controlled and monitored by multiple thermocouples distributed throughout the area. 2. PRINCIPLE OF HOT ISOSTATIC PRESSING

As shown in Figure 2A, a part placed within the work zone is simultaneously acted upon by heat and pressure. Internal voids, sealed from the surface, experience a pressure differential exceeding the local yield strength of the material at appropriate operating temperatures. Deformation occurs in this region as the internal free surface of rhe void is pressed shut. Metallurgical bonds are formed soon after void collapse. Once at temperature and I I-2 pressure. healing ocuurs rather rapidly.

Need for more cost effective manufacturing to produce wrought superalloy arid titanium products has brought forth the use of HIP to manufacture near net shape fotiin,1 preform.s. (3,4.5)

Most major engine manufacturers foresee HIP incorporated on a larqe scale into turbine disk manufacture. In fact, based on results from HIP powder consolidation experiments, the U. S. Air Force is funding several manufacturing programs on airframe and engine components which will use hot isostatic pressing to save time, conserve material and produce better products.

in the mid 1960s, researchers reasoned that if HIP densities sealed or presintered wrought material and imparts superior msechanical properties, then HIP should favorably affect castings with subsurface porosity, or shrinkage cavities. Provided the microporosity is well distributed, no measurable dimensional changes due to pore closur. should occur during HIP. Preliminary work by GE(6), Buttelle(1), and Howmet(7) confirmed that, indeed, this was the case.

Application to the aerospace investment casting field was obvious, for as turbine needs have led to stronger, more complex alloys and shapes, casting quality did not keep up with these demands.

Chromalloy Division-Oklahoma has used the HIP unit especially for the aerospace industry in developing repairs and rejuvenating jut engine parts which were considered unserviceable a few years ago.

3. RESULTS OF HIP REJUVENATION OF TURBINE COMPONENTS

Reduction in microporosity has a profound influence on the mechanical properties of cast superalloy components.

Representative data (Figures 3, 4 and 5) demonstrate that for stress rupture type applications, HIP processing can improve mean property levels and decregse scatter bands. (6,7). At temperatures above the notch sensitive range, typically 1600 F(870 C) and above, HIP has less effect on the mean property level, but by reducing the porosity, HIP reduces the scatter. Obviously, the thinner the cross section the greater the potential benefits of HIP.

Fatigue properties can be improved by HIP processing as welt. Recent data on B1900 and Nimonic 105 alloy turbine blades are (resented in Figures 5 and 6. Note that both mean and 98% confidence limit are improved through use of HIP.

Engine manufacturers have several ways to capitalize on these developments. First, since HIP reduces property scatter more confidence may be placed in the design, either by raising allowable stress levels or extending overhaul periods. Second, the elimination of porosity greatly reduces the chance for premature failure. Third, by eliminating subsurface defects less rejections will occur during final machining operations as these areas become the external surface.

Successful application of HIP on castings depends on optimally combining the casting parameters, HIP parameters, and subsequent heat treatments, to bring forth those characteristics most desirable for the given application.

Use of HIP to rejuvenate damaged components, or extend the service of life limited components, is an area of keen interest to the Chromalloy organization.

Theoretically, HIP should heal internal damage generated during service; surface microcracks may be healed provided a suitable coating is first applied. Immediate angine applications would include turbine blades and disks from both the turbine and compressor.

At Chromalloy Division-Oklahoma, numerous programs employizg HIP recovery of mechanical properties of engine run components have produced positive results. One program demonstrated property recovery of engine run 51900 blades through use 0 of typmcal HiP process conditions followed by resolution beat treatment; HIP at 2175 F(1190 C) 27,000 psi for 2 hourseplus resolution at 2175 F(1190 I), 2 hours rapid cool plus coating cycle at 1975 F(1079 C) for 4 hours plus age at 1650 F(900 C) for 10 hours.

Before and after HIP mechanical properties and microstructures, illustrated in Figures 6 and 7, demonstrate that HIP will improve the stress rupture life and close voids in service run hardware, produce some internal structure homogenization and favorably influence carbide formation and distribution,

Stress rupture properties of the engine run 131900 parts fall on the low end of the new part scatter band. Subjecting the used parts to HIP processing, fully restored stress rupture properties. Later work with chromalloy's advanced HIP unit, having rapid cooling capability, produced equivalent results without the need for a resolution treatment. A similar effort undertaken to recover Rene' 100 and SEL-15 alloy turbine blades has resulted in a production process for rejuvenation of turbine blades now used by a major airline and the general aviation fleet.

_iiiI The full procedure includes a tip weld to restore length, a hot reform to reset twist and warp angles of the airfoil, and hot isostatic pressing to insure that the overhauled part has optimum stress rupture properties.

Initially, the Rene* 100 blades were MIP processed in units without heating or cooling flexibility and received post-resolution treatment as part of the process. The Chromalloy unit, offering rapid quench capability, allowed elimination o; this heat treatment with no loss in properties. In fact, significant improvement in 1400 F(760 C) rupture characteristics are observed. The importance of thermal cycle flexibility in the HIP cycle is illustrated by results of the program to improve the SEL-15 version of theblale. Conventional HIP followed by post-HIP resolution treatments caused severe degradation of the alloy properties while a modified cycle in a unit using rapid quenching improved the alloy properties (Figures 8 and 9).

Thus far, we have indicated benefits of HIP in recovery of stress rupture properties. Similar benefits cai be gained in fatigue. Figure 10 illustrates a comparison of HIP versus no HIP on a cast SEL alloy component. The parts not HIP processed were given a conventional solution treatment in an effort to recover mechanical properties. It is evident that a dramatic improvement of fatigue properties could be achieved by usinqi HIP rather than conventional heat treatment.

Data on Rene' 80 alloy turbine blades reinforce the notion of correctly tailorinq the process cycle to the alloy and component. Chromalloy conducted a number of cycles aimed at optimizing the process cycle. HIP followed by normal coating and aging heat treatments optimized stress rupture properties.

Further improvement was not possible through resolution treatment after HIP and caused deterioration of properties (Figures II and 12). Recently, Chromalloy studied the feasibility of incorporating HIP rejuvenation in the repair scheme of low pressure first stage turbine blades made from Nimonic 105 material. The total enoine time on the blades was 1,290 hours. Three material conditions were studied: (i) "As received" to document the degradation of material propertics;

(ii) Simulated Conventional Repair Treatment (CRT) representing normal repair procedures;

(iii) HIP rejuvenation.

Seventeen blades were used for each of the three material conditions. For the 'as received" condition, no additional treatment wasogiven and the blades were tested as such. •neR simulated CRT consisted of 1900 F(1038 C) for 0.5 hours, air cool followed by A2 9 0 F(700 C) for 16 hours and air cool. The third group of samples was HIP'ed: 1975,F(1088 C) for 4 hours at 28 Ssi fo~lowed by a rapid cool and a heat treatment at 1560 F(850 C) for 24 hours and 1300 F(700 C) for 1 hours. Three blades were used for e evateg temperature tensile testing at 1200°F(650 C) and stress rupture testing at 1200 F(650 C), 118.7 ksi was performed on seven blade test specimen for each condition. The configuration of test specimens for both testings is shown in Figure 13. 0 0 Fatigue testing was performed at 1000 F(538 c) and 150 ksi stress level on seven test specimens for each condition.

The blades used for stress rupture testing were also employed for metallographic examanation. As a result of these tests we found that the tensile properties at 1200 F (600 C) were comparably similar for "as received" and HIP'ed conditions. The CRT material showed lower strength levels, but better ductility. However, for all three conditions, tensile properties were within the alloy specifications. The stress rupture data is presented in Figure 14. The data show good stress rupture capability for all three conditions (Alloy Digest data at 650 C, 118.7 ksi, indicates a 50-hours life). CRT material and HIP'ed blades show some marginal improvement in the stress rupture life over "as received" condition. HIP'ed blades show a wider spread in stress rupture lives although the spread in data is evident for all the three conditions. The fatigue results (Figure 15) show a distinct improvement for the HIP'ed condition; both log mean and 98% limit cyclic lives are superior to those exhibited by "as received" and CRT material. The "as received" material shows a definite degradation in fatigue performance at the engine run blades. As a result of these tests, HIP was incorporated in the repair scheme a;., the blades, life limited before, are used again for at least one more overhaul cycle.

Encouraged by these results, Chromalloy is working toward applying HIP for recovery to other wrought alloys .ncluding Nimonic BOA, Inco 901 and 12% chromium steel. We feel that directionally solidified alloys will respond to HIP process treatments as well. Optimization can be achieved either through resolutionizing after HIP or by direct quenching from the HIP temperature.

v ikI - I . I---0i4 W ''-4

For overhaul work Chromailoy xeconuesnds a direct quench method to avoid potential deleterious effects due to remnant aluminide coating which may occ 8 r dur~ng a conventional post-HIP resolution treatment. Stress rupture data at 1800 F(982 C) and fatigue data at 1400 F(760 C) illustrate potential improvements for HIP of D.S. hardware (Figure 16). Extending these tihermoachanical technuiques to othe:• jet engine hardware, such as disks, will require deveiopinq further techniques in order to maintain dimensional tolerances and to heal surface related defects. The preceding examples, however, emphasize several points.

First, HIP can be used to recover life limited components at overhaul if damage has not progressed to the point of incipient failure. Second, HIP can be used to optimize both new and used part properties. Third, both cast and wrought components will respont. to HIP processing. Fourth, further studies are required in order to optimize HIP cycles for those materials that do not respond favorably to HIP treatment using the parameters established for conventional heat treatment procedures.

RiFE RENC ES

1. Hodge, L.S. Jtc --.. ';* Indue;trial &Engineering Chemistry, Boyer, C.B. Vol.54 Jan., 1962, pp30-35. Orcutt, F.D.

2. Hanes, H.D. -- :s "... , t . .i-..".. Sampe Quarterly, Jan., 1974, ppl-9.

3. Williams, V.A. " ' ;-::::'-. Production, Oct., 1975, pp66-71.

4. Gurganus. T.D. ..* 7::s-:~. -. .. ..' Metal Progress, Vol0106, No.3 Aug., 1974, pp 8 3- 8 4, S6, 88.

5. Dumnord, T.C •r '- ; . .'., ',,-: .. , e. Iron Age, Nov., 1975, pp43-46.

6. Wasielewski, G.l-. . Superallo.s Linblad, N.R. Processing, Proceedtnzs of the Second International Conference, AIME, 1972, ppDI-D24.

7. Kenton, D.J. - ..- ,;':• 7;.: ,':_, .::L2'," ••': :: ;: Diesel and Gas Turbine Progress, March 1976, pp36- 3 7 . * "-2

C14

------~1 1-6

STARTI

AT "RUSC;PE AND TLMPREATUPF P•E&$,SSQ[ DIF FERENCI- COLLAPSES VOID E)AGGERATED VTAUME CHANGE

METAL ATOMS MIGRATE ACROSS THE COLLAPSED BOUNDARY MAKING IT PART OF ITS IMMEDIATE SURHOUN DINGS

Fig. 2A Principle of Hot isostatic Pressing 11-7

a IL ILI -e --. U •

.- i n o - - 0

_ _- I 0

sunow aill

o 0 IV4 z xroo 0 a. IJV;___ __e I 0 - -

0 CV.

10,ac

S2~~ ~~ ~ i 2.•i , Il l ... Effect of HIP on 140OF Fatigue Effect of HIP on Mechanical Properties of a cast B1900 Turbine Blade Properties of Service Run B 1900 Turbine Blades.

4040-- - i STRESS- RUPTURE

14 1400F 94 KS1 so 180F •29 KS

140 70

120 .

0 w 90 sol 401 -0 0 wx~0 .. 700 - " - - .- ilii ----- " - , 600-- -

400 -0 0 giO - -E-- g'

200 20

100

0 20

100

0 Q 4 0 10•OK-- - , o LL 2 ------: ( 3 u0--A-- --j 0 L

"NO HIP HIP NO HIP HIP .4 .+ PRIO1 ENGINE SERVICE 0 NEW PART OATA

"Fig. 6

K = 1.0 A - 0.55 I - .33 H, MAX, STRUG IIIKS1

Fig. 5 - •~ ~ ~ ~ - .-J ,,.- ¾ ...... •e

,i_"v -: * . .. ',•.,..;. •).. ' ;f~x i':,I .- 29

• •, ,...... •. "..t

- ,,- . ,

1 -- . . .-'.!,.,• :-" "

...... "'4• 2. :. " '"

... . ~~~~' t . ,V '4, . .

B Simulated. Coat & Age 10OO X •]!!]A. Illt Simulated Coat & Age 200 X

S t

*4) .5.,lifbi

"At--,c• Fig,.? Cast B 1900 Tete • •, • ,,.,. .. ,.•:•..: .... •-'"• . •t_. -," -•, ••:.• ,. •2, Age 91000 ip ult&Coat Ag& e 200X B.HSimuatdCo oat4- i i i C•.:-. . - :~~~4• .R:•::• ., ., , : .

S1C. Hip &Coot,+Age 200 X D. Hip +Coat,+Age 990 x

l•.•Fig. i 7 Cost B. 1900 Turbine Blade Micro Structure Before And After Hip 14OF 1 85 KSI Stress Rupture Performance

CHAOMALLOY DIVISION OKLAHOMA

d140F 85 KSk STRESS RUP URE PFRFOR VANCE

REFURSISHE RENE 100 Tt ABINE BLADE

NO HIP --- ___

CONV HIP

COo HIP

S.ELI, BLADE REFL RAISHED

NO HIP

CONV HIP

COO HIP

LIFE HOURS

Fig. 9

1400 F 1 85 KSI Stress Rupture Performance

CHROMALLOY DIVISION OKLAHOMA

14DOFP05 h 51IS RE 5 5 I EPE FORM NCE

"EFUf DISH bDUR ,N4 LADE

- CON HIP A NOW IV COO i1P

PROBABIL ITY

Fig. 8 S.. l-l I

ULFE, HOURS

ENGINE RUN -

NEW - _ _ __ _ HIP

ENGINE RUN

NEW HIP. RESOLUTON _ ____I_ _ _ _ _ HIP1

HIP & RESOLUTION I

Fig. 1 RENE 80 Turbhne Blade Stress Rupture Data

Cast SEL-Turbine Component FatigueSIOOOF

3.00 1000~

C., MAX STRE.S AMPL ' ROKSI ' FREQUENCY 331-11 A 095 '. Kt I0t * HIP * CONy HT TREAT

SPECIMENS MACHINED FROM PARTS R4ETIREDPRMSRVICE

0.1 .5 .2 5 10 20 30 40 so 10 0 so go 96 111O WE.9

PROBABILITY %

Fig. 10

4"1 ot %

S Ak !V, t :-W- A

390 X 2700 X A. Engine Run & Conventional Process B,Engine Run & Conventional Process

.54 t 'kt

C. Engine Run + Hip Process 1O

Fig. 12 Effect of Hip on Engine Run Rene 80 Turbine Blades 11-13

!III

I'1~

I tI II

230

110 DIA 150 DIA 10-ý2 TIID iTYP

550 - TVP

M 4 r (TYP I

Fig. 13 Configuration of Test Specimen LP-1 Blades 11-14

STRESS RUPTURE 12W F, 118.7 KSi

uj 100

IL cr: D

•IIII

wi0 w z C) w CC

LP1 BLADE - HIP REJUVENATION

-ma ll Fig. 14 11-15 -

8000 :000

6000

40X

.1000

1000

T FATIGUE RESULTS 0NIMONIC 105

0(8 T 3000F 3(m) -Max- Stress 150 KSi A 0.95 Kt 1 AV~ 1HZ

095Log Mean 00I wA 90 98% Limit

60 Kt 1

I w

t.U <, z 0

LP1 - HIP REJUVENATION

Fig. 15 Effect of HIP on Mechanical Properties of DS MARM-200 Alloy Components

i4G0F iATIGUE I,OF 2SKS1 240 - STRES&RUPTURE

FLAT SPECIMEN4S 30,000 220

10.000 FLAT 20 -SPECIMENS T 140OF A 0.9s I-i o Kt1 1.0 MAX.STRESS= 0

130 KSI - __ U

1.000 IGO

140

BARS ARE OBSERVED BARS ARE OBSERVED RANGEOF DATA RANGE OF DATA

L I

+*+

,a04 0 4 . F 0 . 0 zt' U+0(auWO

Fig. 16 RFCORDER'S REPORT SESSION I1

.A..A.M on Nationial Acrospac L[aloratory N LR P.O. Bo\ 153 8300 AD i-inieloord. Net herlands

File pa pem in the session "L.ife ¢c\tc ,ion and repair IIF cot d he brougtL unde1' three difltrent IlIleading,,.

I Some papers. the papers whiiCh were presented byvh %I arMisnsi otl libar and %Ir Pichoir of ON F RA. LI.d1t %Nitih high tenmper¢arture coatirgs, their prote cti on and the effltco of coa uing, ol h mciahtni-,ci propcrtiC01tit he substrfate.

_21 Te paper giuen by Mr I lollotrat of Snecima cdn Ibe brought under the eading: Repair braic prioceL,sc\,

3 i lhe other tuo papers. given by Mr Leach of AWestingltouso antnd Mr t.IVet of (kiirimualt , disciussed tie reiu er :tion of turbin e conintllelll 11) tile Use ot" !Il IIIIP prncess. wihich 1ireans%I ot Isotatic Prcsing.

Although all tile above processes do not seeIn diect!. ,Ia-etltd It' Ceach other. they have sw,,¢ral thinils in coinmon. Firstt). they\.r IeMostly applied 0on IhighIi tnmporal treIturbinic parts. Secondly. :ind imos imstportantly- thle all cr\e the Salinc purpose¢ tile extensioin Ltlservice life of1 high temperature pmrts which arL Very. not to say extremely. e\petisiie. To give an example: present 1st stage turbine b[lde' will cost lip to about 2000 do+llars c•ach: ncvsegmneiit prices can ble een higher.

Let us go back Inow to tie first s lbicet in thi s session tile, coaotiilgs. Nit NIarijnissel of t" bar told utii.llout [lie IHkil 3(00 coating. which is a li-Si coatingn dc& eolpeti ty I is coin1panlN for Ue (1ii indLIstrial gI. t i rbi iiiL,. 1 he Coathllg is oh tah sled by firstlyIpplyiiig a thinl itinitm Iniyer on tihe surface b\ uneans | iofii-pltatig. followed Iby ai kLtiusion hleat treatentici. Subsequently tile part is pac k-silieoniiZe•i and flially aged Tile coating aptplication hiet treatiii'n| i; part of tie heat treatenl+2tt of the sublstrate. According to tile speaker this Ti-Si coating should demonstriatc i superior corrosion resistance at atround 70U0(' and eq.il al Corrosion resistance at about 1O00'(" in conliparison to the Ml rAlY type ovcra Ns. UnlfottLinat e.) these comparatise results were not included in tile paper. Nevertheless according to tile published np testing results this coating seems tol he fairly promising. at least for the lower tempcrature range.

Mr Pichoir tif ONU RA gine a presentltioni about the eff.ect of Coiatilngs oii tile mechanical properties Ofl siperalloy comiponents. The adverse effects of Coatings can be due to ti I tile thermal cycle of the Coating applicationi treatment: 12) the reduction ii cross section and tile coating weight, resutltintg in higher loading of tile net. toad-bearing cross sec tion (3) the interdiffltsion between coatin, ,inld substrateLduring serice. Which miehi t result in uillfavol ratitL precipitation :and 14) the niicrost rictuLre of the Co:tiiig itself. e.g. a ltow ductility of tile coating resilting in carl\ crack initiation.

Tlhe author cnmphasised that dependent upon tile nature of the coatliig t pack-aluniinides in oi'oparisoll to tile overlatys) tle effect on the mechaniical properties cali be qLuite diflerent. E.g.. tie O ierl.y cIoatings show less negative et'fects on tile mechanical properties because their thermal application cycte can easil. be illad¢islec patile \%ith tile superalloy substrate heat treatment. Furthermore. the interdiffusion between overlaks aiid utistrale is omily of( iniii0r importince owing to tile better stability of the overlay coating. Finadl tile o•erlax shL•ws nt.tielh be[tter ductiliy.

in tihe discussion on these pa11 rs the questtions were especially LIdirected it tile dutctility moftile Ilcoa t 3,00 coating aild the behaviour under thermal Cycliig Acý ordiing to the author (lie ductility of the coating and its vdIic be h•aiour Shouild be good.

The second subject in this session was repair hrazing.

NIr liloiorati gave aniuicit iing oaiecli ul ilttionab the iq tteciitfd ifitiston bra!ihi g for re ptii.:it0-r;iLkckt'i m|lid even further LIgraitletl tiulrine stator coniponents. lie described tiso processes: I I brasag•e diffiusioi, which is a diffusion brazing process isinigai low melting point braie fille mietal. based on aiNi-(Co or Ni-('r base alloy with Si and IB-addi0io115. fle brit¢e filler metaltcan bIe applied in Ihhe forit of a nilcat I'foil or ulsa pI•ste co•illainiing braze inc tlal po\-tder. This process Lanlbridge gaps tip to a maxiiiimum of abouit 100 pll R2-2

The ••cond prkxeos thie autlior mentioned wu -rlchargcmenit bt . diffumionl Which iN Ulo a diftusiogi braing process. |lowcver. in this Case tih braze filler metal is always applied as a paste. Furthermnore. apart f'rom tihe hOw InetinI- point braze osMIdt'cr a second INpe powder is added. ihch hIias TIie a•;liekor allpros•Tili at [ii z rle •'l[•nop o t Iiiis ithe bae material of the component, aad which will not mell during the proe-sks. This allows for the repair o'f wide siaps anti even allow,; for the build up of worn-offand corroded sUrtacs. Thi deseCrihd teclhniq tics. whIlch ,clln To N, vcrv similar to the AI)II pr:ocess d-veloped 1y (CF. look Ncr' proniiising

Tlie final subject in this .ssioir wa,;s the potential ol IlllP proscossingio reoins ncii; iturlil_n :totlrporenill1. BIlth atlthors. Mr Le ach ofV' ,elinvhouruseandljn r %d-Vet if ('honialloh pt'iited It the en+.lefri~il e CH, I of t e ,onbinrjtiont (it applied lprcsstrre anrd bleal treatament Jhiring thi"llHIP proce&,,

l)uring hippingg tle coinponenl is simuhtaneouioly subjected toi high te'n |lrat tire and xerx hi.i=h prcs,,irc Such a tre•_itl- mieIii will on the one hand result in limninating structural disb-oitinroit, i¢. . .\ eh'iing upli intenial cree p s oidsi. On•I the other hand the proccss has the potential of reheat treating the compoient and rL! ersivtg the microtructural dcnIrada+.oni ""hich has occurred in senrice. On Ce\eral alloys hllyirpi has show\n really reiniarkahlc re, Itit. other allo. , hipping .cents to ha€e a niegat i\ etleet. Tile iii os important p ob1.ll ,still i, the selection kit a Nuitul e heat treatnhinI r c),le fior each indiidiral allo\. lecari sc lemperatjtire has a mnarked eft'cci on the diNoftlutioin )It the -N' pha,,e inrd othlir precipitate andili1 grain ik oarselllnng. AI:o.. the cooling rate at the e nd of the fIll' prces ileItl or drining re•li'at -t rc tvlien|t ccc'c is a ll ur fact or. A iiore" tAud aIrridalllt.'tl knoulcLe of tile coiir lhi ied e tII'1 oi' hi'.it aiLd presure s•ec, i tol te iiec-ars lk, -diii tle full potential from this very promising tc•,.'ntologv.

To Coicluide: New%Coating!, aire being de\eloped: man\ repair procuesses are appearinv. e.g dilftfsion bra/ingi mitl the optioi of adding e\tra pa.reint me.tal pois sdcr tI' bridge broad rackN.r or cien til ldild up aitfoil sir';rfa'C, a.lL hllip in i o-ndr'erka% ax a means ito rjuieniate rejcted lurbinie part• ,,ill wartirlr iii stock ftor betctr tillei, All tihre, procesIse are ihighl' mi portaint. vet I iiiIissedL somelin k\gihi17, hardLyiv 55: spoker itLLduring the prceclnttiioLni ailI diossuioti,, I x%ill i melletion t\\o simple Ilhillgn,:

1 1 , there an> ody, lihere Lvho can tell me how kol :lanldetcriiilic tile rcminjiii g life of a coatiing wIilii a bllade Co;ie1s ini for ove-rhaul? I will intake tltis clear: it a blade comes in ailr it still looks good. whto is toingi.i•sax Ithat iil cain.111retirn it to s•n ice inStead Of r-cCoatlinc it'" caus it tire coatini! i, goinni to fail ,liiirtl% fiter reti rniig it to se nice thlen t[lie blade will not kcxcii be rlpai rabl I at tire i\i Li'rhtaiio+t orvini to se+'-crc. pawrit ruetial attack. ]hus lieth+retmaining! coatring Iil+: has Iti be i.reasured in soime•- •a).

AI-1l l repair processes like diffusitnon bra? it andalso hipnping ealt onl\ eli iniiinut ..urf'ac¢ coniictcd cfacks wvith Vcr-y Clea1n slurfakoes. Htenc cc lcair ling is a imialr ritpie. , Perhaps e cair lix.,er t lieset Lopcsliiidtrigp ti

H. S. *xler

S--he elfettivene:s -f h t i . *:tnic s: :=~.'.eCsiE,:•'i: "-.::i t. •=a. S.•-•...•'•.. ~~accsaazuioter• during osceer -'in' e.se inves~ic-ted ?teinsec.e. , -i:. ut.[ i •,f ~~~~~~t4.e2Xi-rtru are life,•F.--_ti,")-h r-. .pii, ::issntetce=- te. ec.tirnui::g tue creep test, -_e-- '-e--niiF£,t:::•]r. Ac..>i : ct -r rerties f-''ea tese t.'r r:;tr intr.-~iteiht an :-tral:te :r•te after h.prirg at '-ome s-tage ri ecnrvTtetr/ 'e-ha .- • '. ir.emtt O O ; i.thc r"-pt-r lie' aheao he i:e ' h 'eee'•i. Fo"tit'-t=~e~ t''art ''tte o'ree- re-'i.•t-- 7e -- a's e--r'-h r•tc!i.•i:•i:.', .e ~'

e'-eat-reci! an'o-- i-h -res--ure, cal ci:ii a-nti,' cae ::r{,.•t-itcz' -•e'_ _,: ti .e! e'tauliashei fo' wrought supe-a vit' th-t'f-•e~er-e -;raelznea- ne'- .cc •-'-- -a ::s'.'-i etf:

Sprocess' t' re'-vae this in-te-rnal tree- -; ---- trer-ek res'tor ~i-g yrepee :'-e -Ia -x:e:-m:t:ae . pie~~~CeCsan , '-rooiat- subjcct to creep def._r ¼r ntitna ¢:nire tth iini.,]eie c:. -c-eooee t a .in~.ificcant extent heueu 1',"•ie tuu'- eFeddrh 4ta-reset is-kx,: ~~e Sott ;vestia~te t-e feas-ib-i ity- 'f '--iag the sar 2•'r-esL ore•:rIFri tn- i -'-1i...eni- cast auperallan, t.7s .k whi h was"}'.s.• tfail as -are- l -: aity ¢:ratiU-p .. Fei.u•wok: this sal~o a. sh. t:atftr, tiarious ti:,es fepsr nc-e oiin ?- h.Fit trametsis gave amore repraodu.cie, vales of. rupture life fr rc'i -nc -te~wit' .taS.-trio:~~fcn m: e merat in propertie-s (,r- -arcud result xi inc'rese rupture lives- u,. S. ai~t•[ ., ••etsi solved relatively snort live'- to ru-t-u'e -omparedtc comp..ents insrieadA;etr : h r:e work nas ceer. th-at 1las .-tr-c--'e-a *ere F.=e t ive sive t. ru'-tu'e f-','i.celz"<,in i.'• :. irportant to deter'sne the lii-feZont ofcesJa'.g7db •-• .. i-•d• ,:-'::t,-•:r hipping recame i-n'effetiv.e, cre tes-t- ha3ve Peen inte'rrte at v-ri,:'- '-ages- "'ri-i '- ruer.

Two batches of testyicces were used, identified respc'-vcl• •s½'--: 1' h r -n 'e'i'e of ' , EQS series were machined from investment-cast blanks -while tros-e of the at seie were rac- iend 'r-' cast "carrots". The SCI specimens had a more uniform and sligntl- larger grai .. ie cut extensire test'- '--- -ed that The difference was not sufficient to affect the dreer reI~rfon•nce '.1C' "emaa~ei -a'a\-• h • used in this worka was within the limits for the nzormal coe, e: cia e'ecii' 1atl •• h .Hr•. a 0 rent of 2.h/i 20 CiAc + 241-v/35 0e/AO was applied.

high sensitivity; creep tests were carried out at 25.-'-° using re.t-tieces of t; eli• ie~-:s parallel portion. SOna.•; gauge length, 50.5•mm; gauge dianeter, ...... li•uge Ia g' - ws-a t:mi.. k-" circumferential ridges• to which extens•oreter linbse wera- co". ectea At- te-t.at' ' .'--*e CarrieI out on EQSHmaterial and those at 2'-Cbt~aon SB material. Tostdy tie c e'-et -. :.4i••= •e'~ .it!- vary'ing amounts of creep damage tests on different specimens were termnint'e• ¼fter y-'--g'-eei-"Jan; ut• creep vi:. from early secondary to the tertiary stage of de'frrs~txon * .e te- trieoe' "ere te:f~nT regenerative treatment which involved hipping for Sh at '!CC •a aproeut :"aon -.--f:'--'biss fr e, _ __ b::~~tthe commercial heat treatrent. The surfaces were cleansa ¼ reoi-gai'-'.- .'-'•,". f!' t--e r.ee-- and each specimen was then rotested to failure uradvr the sa- 7-anito f 4t4's-:ata~a'r • -- previously,' base-line data were obtained by. testing two s'-eclaim-'- t•i{45.a teein .•'u , r•t I •-" • ~~without interruption and similarly one specimen of Sf: m'terial •t2' Fa

•---• ~Tests on. seven: specimens of IBM at 1i?•Paa were inter-ru'-ted -at tUe ti-me- a'd -. vt''-in det-ieii. ! !! ~~~~~fable Iand described as condition 'P * After- ap~ivirg the re-aceati'.e- tre½ ...-at ~ C test -,fie III .soeciens were continued to failure, bttwo speal- ..... vi-. a, .. aiF ret-ni-c -.... •.• ~graphic exaninstion with no further creep tes-tin. Pita I-a- t'ic -espei'r -. St &teria- whic 'eo~ie-ae III ~~~~~thesare senuence of testing, reg~eneration ann r-eteutire, but un _-an -aplied ,a-eec strea . ,-r'bl,:

condition 'C. is detailed also in Table 1 and the combimed lile, ise incousiv ol that iefore ano sfter the hipping treatment, is given by 'S• + C'. The total life to i--at'riF •, ie - rrared with th'r f.or teati- -- II~~m to fracture without interruption in the same test conditions, ic. cor-diti- 'a' in Cable ',, and it w1l b'e observed that fur all tests except that stopped at the earliest stage ."•{-r- th~ere was an inca-'Ise in. total rupture life and, to varying extents, in r'uptur'e luctilit; as a re ault the regenerative treat':nt. I ~~~The significance of these res'ults becomes sore apparent when the our-ati'i .A" th:e irniti-al ca-ce- -ri

to hipping, express-cd as a ]crcentage ot the expected nominal life, isitted -.5 a £unoti:n c-f to'et~t'il S12-21

life obttadnd by crt-r-bing the dirrstt'r- In creep prier cAn mubmepq2ret tm hirping, alsO expressed "s a percentage , f the expected nomiral life. A graph illustrating the relationship obtained is shown in Fig.1, and it is evident that a greater tenefit in increased lite was obtained for sp'imenn tested to tic- later stages -f creep 1ri i t .its :.,. 1xaFple:- of" the crlee carver. -4tFi:c-6 ' !ore e!c- flter t.t- ro:Cer.- erative treatment are given in Figures 2 and 3, for tests interrupted during steedy-state and tertiary creep respectively. in tig.3 a creep curve tram a specimen tested to rupt'u'e witrout irterruption iL also shown for compari.o. g aX 3 and the data in Table 1 _hiw tiat tze crce.}j tot we:C rTI- stored by the hippz:::g p -;d nricef-o: .- ig.4 it in evident that the greateAt --.v ci t i cree rtea- t ance was obtained , the specimens with the highest creep rate prior to. hipping.

The re:silti of experi-.ints rteid elsewhere (ý 1 s howed that cav ties .-i. train-. Lunarie. ir. t li.5AoL0 were observed 'after .V'Ih in tes.ts, carried out at -', 'C snao - *, £ * thiat -i d,!-gc.j boundaries would be expected to occur in all the tests performed in this stuay. A metal 1 graphic evxeii- ation tc investigate the effect if hipping on creep damage on all specimens was clearly imposFible because of the lestructive nature of the examination required, so that observations of the effect Pf hipping have been confined to two representative cases. Thus tests on two specimens, ACtSHD ans /CIV, were t ,- at -xpproximately 4500h and on a further two specimens, ?IvC1IllD and ?7XlHloD, at approximately 75C00h (Table V . Subsequent to the application of the hipping procedure to these two pairs of specimens, one of each wasi re- tested in creep while the other was sectioned for metallographic examination. There was no evidence of any intergranular cavitation or cracking in eitner specimen so that it i. reasonable to assume that the creep damage haid been healed as a result of the hipping treatment. ~liSCPD'SSiofl

It is evijent from the results pre.s-nted that hipping can restore toe creep perforsance of specimenr: crept to esjrious stages of secondary and tertiary creep, presurnably as a recalt of the restors,tion of thy original microstructural condition (6,q9. Except for the specimen hipaed after less than 2tulof the expected life, the total life to rupture was extended in everv care and the increase in duration o;ar nore marked for the specimens which had been crept for the longest tiUes prior t. reveiving the regenerative treatment. A feature of the reaults is that for the three tests interrnpted after creep durations of between 3CYX)and LaitCh (creep strains of up to 1.194) the life to rupture after hipping was approximately the same ie. This''000L. may represent the maximum life recoverable by a single regenerative treatment. Significantly, for the single specimen crept to the beginiing Af rapid tertiary (2.6A, strain;, the life after hipping was approximately 1 beth less than this apparently optimum recvira'le life. Thur the 1 average line drawn through the points in lig.1, w ich assunes that there is no decrease in the efficiency of hipping at high prior-strains, may not be valid. However, further wnrk would be necessary to establich the precise form of this relationship which could have important implications for the practical application of hipping techniques to the recovery of propertie.: in components removed from service.

There was no metallographic evidence to suggest that hipping became less effective after higher creep strains since grain boundary cavitation and crackint, wqs removed in the specimen crept to the tertiary stage. Also the creep recistanre was reo:t-:red suggestin:6 that the morpho,_!:gy of the micros.tructure wan similar to that in the "as heat-treated" condition.

Surprisingly the recovery of creep life in the specimen interrupted in the early stage of creep (c.tv strain) also showed less than the apparent ranximus amount of recoverable life. ITis is illustrated by the dotted line in Fig.1 since it would be expected that any benefit of hipping after creep would result in a total lile to rupture of more than 100$. One possible explanati'rn is that the hipping operation was somewhat damaging. In this case alac further work would be necessary to establish a complete unrerstanding of the behaviour since in a limited programme the effects of variability of performance, characteristic of cast alloys, cannot be assessed.

Despite these limitations the work has shown, for the first time to our knowledge for a cast alloy under the conditions of test described tvi.. low stress and long durations), that creep performance con be restored by a hipping treatment. In terms of the practical application the evidence that significant extension of life can be obtained by regenerative treatment applit! in the later stages of creep deformation is obvioortly advantageous particularly from the point of view of planning maintenance intervals. However it will be important to identify the stage at which effective regeneration will not be achieved. For the present alloy, comparison with other work (10) suggests that full regeneration of properties is obtained up to the stage at which intergranular cracks are no greater then about 0.03mm in length. T-hus, for this particular alloy, the observation of internal crack length may provide a simple criterion for assessing the probable efficiency of a regenerative treatment although parameters appropriate to the lower applied strosees encountered in practice may have to be established. Also no effort has bees made to investigate the effects of successive hipping operations but the results from work on a wrought alloy suggest that further restoration of properties could be expected (11).

Ln the wider context of repair of components the use of a welding process may result in a locally heterogeneous microstructure and occasionally in the formation of small defects. The application of regenerative treatments which include hipping after weld repair will ensure the re-establishment of a microstructure with good creep resistance and rupture properties.

CONCLUSIONS

The regenerative treatment referred to in these conclusions is a combined process which involves hot isostatic pressing followed by commercial heat treatment end tne conclusions relate specifically to the cast superalloy IN738?I.

1, The rigenerative treatment was effective in recovering the rupture properties when up to 8,5% of the nominal life had been consumed.

2. The later the stage in creep at which the regenerative treatment was applied, the greater was the increase in total life to rupture. A maximum improvement of 55% in rupture life was obtained. 12-3

3. The maximum efficiency of the regeneration treatirmnt, in terms of subsequent added creep life, is approximately constant over a prescribe! but wide range of creep duration.

4- The creep resiotance was restored for material in which the deformation, had Lee-- allowed to r':, well into the tertiar.y, stage uf creep.

The Ms.imom ereLp rate after the regenerative treatment tended to decrease as the creep rate at awita the test was stopped prior to the regenerative treatment increased.

t. The practical iaC toe ion' of the work relate to the use;e of rei-enerativp treatments to extend the life of gas tornii rorpoen''iurevd from Lervice druinr periodic ovenrlaal.

REFERENCES

1. U van Drunen, 3 Liburdi, W Wallace and T Terada. "Hot iaostatic prýocessing of 11-/3S bladts". jProc. of Cent, on Advanced Paui-x*ton Proces-ses, Florence, Italy. Sept. 197b. AGARL Cr 256. Paper 13.

2. G E Wasielewski and N . Lindblad. "Elimination of Casting Defect- using HIP". Proc. of 2nd Conf on Pruocessing of Superallovy,, Metals and Ceramics Informatiui, Cxtrr, Battelle, C luambus, c-isc. 19:2. pp. 1 i-DZU,,

5. J G beddoes and W Wallace. "'eat-treatment of hot isoFtatically processed I; 73t investment castings". Metallography 1950. vol. 15. pp '65-%a.

4. N A Stevens and P E J Flewitt. "Removal of casting poroEity in superalloy; IN 738 using hydro- static pressure". Met Sci 100L. vol. i4.' 81-88.

5. H R Tipler, Y Lindblom and J H Davidson. "Damage accumulation and fracture in creep of Ji-baoe alloys". Proc. of Conf. on "High temperature alloys for gas turbines", Liege, Sept. 1978. Appl. Sci. Publrs Ltd. Barking. pp 359-407.

6. Y Lindblom. "Maintenance techniques for increasing the service life of rotating gas turbine components". Lund Institute of Technology, Lund, Sweden 1979. pp 71-134.

7.GC van rnen and J Liurdi. "Rejuvenation of used turbine blades by hot isottatic pressing". 6th Turbo-machinery Conf. Houston, Texas, 1972. Reprint H-268.

8. H R Tipler. "Metallographic aspects of creep fracture in a cast Ni-Cr-base alloy". Proc. of ICF5, Cannes, April, 1981. Adv. in Fracture Research, Pergamon Press, Oxford. 1961 vol. 4. pp 1587-1594.

9. R A Stevens and F E 3 Flewitt. "Regenerative heat treatments for the extension of the creep life of the superalloy ID 738". Proc. of I.C.S.M.A.5, Aachen, August 1979. vol. 1. pp 439-444.

10. H R Tipler and M S Peck. "The creep fracture process in a cast Ni-Cr-base alloy". National Physical Laboratory, Teddington. Report DMA WA) 33. June 1981.

11. Y Lindblom and H R Tipler. Unpublished work.

ACKNOWLEDGEM34TS

This work formed part of the European COST 50 Project on Materials for Gas Turbine6. The hipping was carried out at FEV Linkoping. Sweden under the supervision of Mr. Y Lindblom who also provided valuable detailed advice and information. Assistance in the creep teeting by Miss L. V. Game and in the metallographic work by Mrs M. S. Peck is also gratefully acknowledged. .• 12-4

0 Influence of hipping treatment on creep properties of .102381, at 850 C

Creep rate, 4 Dctility Applied % per h x 10 Stress Condition Mark Duration IMPa Minimum End of h Elongation RA% Stage B

17o B 7DCH15D 3.3 3.3 1505-S 0.5 0.6 C 71DCH5D 2.5 5758-F 7.4 9.0

" 5 K ,'DC!ISD 7263 7.9 9.6

1b, 5 7DC1i6D 2.9 2.) 3005-S 0.8 1.0

C 700LC16D 2.7 - 7052-F 6.0 11.0 B + C 7?WIf6D 10057 6.8 12.0

V70 B 7?LCH7D 2.26 2.9 4697-s 1.0 0.6 iC '?7PCt7 ,2.6,j - 7305-F 8. 1.0 13 + G 70007017 12002 9.0 11.6

170 B 7DC1180 2.14 2.2 4518-S 0.95. 0.4

170 B 700Ii9D 2.37 4.5 6045-S 1.5 0.5 C 7D0;09 1.3 - 7,38-F 6.6 11.0 B 4 70000 158 8.1 11.51

B 71)CHI0 2.4 8.5 7555-6 2.6 2.6 C 7DC0IOD 1.7 - 5733-F 4.0 8.9 B + C 7DCHIOD 13288 6.6 11.5

170 B 70DC111) 2.2 9.4 7540-s 3.0 2.2

170 A Mean of 2.35 8680-F 5.25 7.0 Two Tests

250 B SUOl 21.0 30.8 553-8 1.5 1.3 c SU01 23.5 - 742-F 5.o 11.0 B + C 0001 1295 6.5 1,.3-

250 A SU10 24.6 86o-Y 4.5 10.0

F - time to fracture S = test stopped at time given

"Condition A Cht + creep to rupture (Cht ;2h/1120'C/AC + 24h/845OC/AC)

Condition B Cht + partial life creep prior to hipping (hipping 2h1/1180 C/170MPa + Cht)

Condition C - Creep to rupture after hipping 12-5

S901 ... o.

80' a0 esis' al 170 MPa (nominal rupture life87OOh) x lest at 250 MPo (n•minal rupture tle,1,60h)

A Telt% at 27144Hl irwnotrui rupture 9lfe, E70tr- 1100h) data ref cN 60.- A ý A

a/ &

// 40.

30. I,

a / a / -to./ /

080 9o0 o0 1`10 1t0 130 140 t50 t6o Total HiP, IP. befolre and after hippirng.% oi expected nominal lile

Fig. I Extension at rupture lite of IN 3SLC b'y use of a regenerative hipping treatment oiler difterent durations of creep prior to expected nominal rupture in the cotnrnercial heal-treatment condition.

Specimen : 7DCH6O Condition "B':Partial lfe creep test in condIition 0 2hitt2O C:'C .24hIa45sCIAC (Chl Condition 'C'! Retest to rupture alter hipping Q(h11180'CI170 MPo . Cht) F - lest to fracture

6,-

2- Creep test i nterrupted here *8'C

2000 d0oo 6000 6o6 10o00 12001 t).00o Tlmeh Fig 2 Effect of hipping on creept behaviour of IN738LC in lest stopped during steady state creep a 17OMPa and 8504C

ni 12-6

Specimen: 7DCHOB 0 Condition A - 2hIl120tCIAC .24h/I8&5 CIAC (Cht) 8 Specimen: 7DCHIOD Condition 'B' Partial life creep test in condition 2h/1120IAC1 24 h/645°C/AC (Cht) Condition 'C' - Retest to rupture after hipping F (2hl11600 1170MPa *Cht ", 6 - Test to fracture

C 0 0 4-

Creep test interrupted here 2

0 2000 4000 6000 Boo0 10000 12000 14000 Time, h Fig.3 Effect of hipping on creep behoviour at IN738LC in test sopped in tertiary creep prior to rupture at 17OMPo and 850 0C

- 3 CL

22.5

U 1. E

2 4 6 8 10 % per h x 10l Creep rate at interrupted stage(B)

Fig. 4. Effect of creep rate reached at end of initial test on minimum creep rate after hipping treatment, Mii' S13-1 *

Repair and Regeneration of Turbine Blades, Vanes and Discs

H. Huff and J. Wortmann MTU MPnchen 8000 MUnchen 50 West Gemlany

Summary

Even today, the repair of incipiently cracked turbine components is essentially limited to non-rotating parts. Repairs to rotating parts are carried out on low-stressed areas, such as seals, only. In this case, weld build-up has proved to be a suitable process. Stator vanes can be high-temperature brazed following reduction annealing. However, problems are encountered when it comes to making sure of the complete removal of oxides. A highly promising method for increasing the reliability of turbine blades that have been in service lies in their regeneration by heat treatment or HIP-processing. Results to date have been so positive that one may reckon with the use of regenerated blades in the near future. However, a prerequisite for the use of these repair procedures is a guarantee of reliability and a knowledge of the stresses that occur during operation.

Repair of Turbine Blades, Vanes and Discs Turbine components suffer from different kinds of loading, which in the highly stressed parts results in life-limiting consumption by creep, thermal fatigue and low-cycle fatigue. Several repair procedures have become well established for non-rotating parts in particular. Repair of rotating components, discs and blades, for instance, is usually restricted to lowly stressed areas, especially sealing fins, blade tips and couplings (Tables).

VWelding and brazing techniques have been successfully developed even for materials that are difficult to weld, such as Ni-based superalloys with high ' -content. Regen- erative treatments have been specified and will be in use soon. This applies to turbine blades, and investigations have been started for disc materials. In any case quality assurance methods have to be available to attest that the repaired component, base material and zone meet the specific strength requirements. Component tests under near-to-service conditions as well as test runs will in most cases be necessary to validate the repair procedures.

Repair of abraded seals Because of unbalances and during running-in, seals of discs, shafts and blades are usually abraded, resulting in a decrease of engine efficiency. Several repair methods have been developed.

Figure 1 shows for example a detail of a repair instruction for turbine blades with seal tips at shroud and platform. These are removed completely by grinding the assembled rotor. A mating ring is prepared, sectioned, joint by brazing and finally the fins are restored (Fig. 2). The brazing temperature is about 1200 C, well above the solution temperature and neur the incipient melting point. Embrittlement of the grain boundaries by borides can also be observed, which may be detrisental to the thermal fatigue behaviour of the fins. Temperatures well above 1000 C have to be used because of poor wetting behaviour of these alloys at lower temperatures (1).

To overcome the problems resulting from the heat treatment of the whole blade at highest temperatures a micro-plasma spraying or welding process would be preferred.

Figure 3 shows an experimental TIG welding, using blade material IN 100 and welding filler material Nimonic 90, which shows a micro-fissure-free heat-affected zone, despite the poor welding properties of IN 100. Thermal fatigue tests proved that this repair procedure produces suffictent properties. Stellite filler materials will further improve the wear properties of the sealing fins. Abraded blade tips can be repaired using the same procedure (Fig. 4) (2).

Vanes

Crack formation in vanes is very common, in particular at trailing edges and near changes in section, owing to thermal fatigue-type loading. Because of the high temperature gas environment these cracks reveal heavily oxidized surfaces. Because of their thermodynamic stability, sophisticated methods have to be applied to reduce the oxides. Soaking under hydrogen or more effective fluoride atmosphere is used (3). S ~13-2

Problems arise particularly fom narrow cracks which have proved difficult to deoxidize and inspect. Incompletely cleaned crack surfaces will not be wetted by the brazing filler and consequently will give rise to crack propagation (Fig. 5). It would seem necessary to monitor the repair process by metallographical examination of a representative sample.

Regeneration of Turbine Blades Problem The remaining creep life of a set of turbine blades has to be assessed on the basis of several microsections taken from samples. The criterion for the degree of consumption of creep life is the creep voiding of the grain boundaries. Classification of the sectioned blades is done with reference to a standard, which is assumed to give a correct correlation between consumption of creep life and the degree of voiding (Fig. 6). The mean creep life damage of the test sample is taken as representative of the whole set and the remaining service time is assessed accordingly. This procedure is unsatisfactory because of - inspecting a relatively small test sample - difficulties in correlating a definite microsection with a standard - taking mean sample creep damage to define the remaining service time despite relatively large scatter. We therefore feel that the probability for premature turbine blade failures associated with this process should be diminished by a suitable method, which could be a regelr- eration of the blades. The blade in question, made of wrought Ni-base alloy Nimonic 108 ;20 Co, 15 Cr, 5 Mo, 5 Al, 1.2 Ti), is cooled and coated (aluminium diffusion coating). An increase in the usable service lifetime of at least 30 % is thought to be sufficient to give an acceptable standard of in-service failure probability. Treatments promising successful regeneration of the creep properties are: - heat treatment typical for this alloy and

- hot isostatic pressing on condition that no loss of other mechanical properties Uf the blade is caused by the treatment. This means that the impact and fatigue properties must be examined closely after regeneration. The investigation has been performed on test bars and blades. The bars were precrept up to about 75 % of their creep life, treated for regeneration and loaded finally to fracture. Blades were taken from engines, treated and tested in a hot gas test facility and compared with both new blades and those that had been in service. In addition, fatigue (HCF), tensile impact tests and metallographic investigations were carried out.

Typical heat treatment The first approach to restore the creep properties of samples and blades was the application of heat treatment typical for Nimonic 108t 4 h/1l50 Or / AC + 16 h/1030 °C / AC + 16 h/700 °C

Results: Creep properties of precrept bars (up to 70 a rupture time) were shown to be completely restorable (Fig. 7). More severely elongated bars ( )7 %) were affected positively, revealing some ten per cent increase in total life. Elimination of creep porosity is achieved by sintering and grain boundary migration during the solution heat treatment. Complete removal of voids was not attained (4). Grain coarsening and precipitation of large inter-metallic phases within the coating occurred simultaneously. HIP Treatment

Among others, the following HIP parameters were tested.

1 h / 1150 'C / 1600 bar 0 I h / 1050 C / 1800 bar 1.5 h / 1070 'C / 1320 bar

0 0 Partial heat treatments at 1030 C and/or 700 C were carried out additionally, the complete heat treatment showing the best results.

Results:

Total rupture life of bars and blades - already damaged in a test rig corresponding to aMout 75 % of their lifetime - was improved by at least 50 %.

No significant difference in the high cycle fatigue behaviour (bending fatigue) of virgin and HIP-treated blades and no embrittlement were observed.

Even subsurface grain boundary cracks can be eliminated (Fig. 8 and 9). No grain boundary cgarsening or degradation of coating were observed at HIP temperatures lower than 1100 C. The standard deviation of untreated blades is stintr. = 75 h, of treated s treat. ' 40 h (Fir. 10).

The microstructure ( 7' and carbide morphology) is restorable, with one peculiarity, that 7y-precipitates after HIP treatment are of cubic modification whereas the normal modification is spherical. (The same is true after an additional heat treatment.) The explanation for this behaviour is probably a change in the misfit of r and 7' under HIP conditions to lower values as well as a change in solubility of the alloying elements.

Conclusion

The creep properties of damaged test bars and bldes proved to be restorable - at least partially - by both typical heat treatment and HIP treatment. But grain coarsening and degradation of the aluminium diffusion coating precludes the application of solution heat treatment.

0 A HIP treatment at a temperature level of about 1050 C and pressures of more than 1000 bar would appear to be a successful method for restoring the service properties of turbine blades for more than 50 % of their designed life time.

Quality assurance has to guarantee closely-kept HIP parameters such us temperature, pressure, cooling rate and contaminants.

The application of a regeneration treatment will only decrease the probability of an engine failure, not prevent it, of course. For there is as yet no reliable method for sorting out those blades which suffer from internal cracks starting from the cooling holes. Even high definition X-ray radiography has not proved suitable. Because the probability of internal cracks of critical sizes is significantly small, these defects would seem to be tolerable.

In 1982 regenerated blades will be fitted in three flight engines, and others will run in an endurance test.

References

1. W. Stoll, Thermal joining of high-temperature-resistant casting materials, "DVS-Bericht 53, 1978, 80 2. H.A. Wilhelm, mtu intern report 1981

3. J. Parmentier, P. Calmels, Brazing stainless steel in a stable reducing atmosphere "of fluoride, Welding Journal, Nov. 1977, 15 4. Hart, Gaytor, Recovery of mechanical properties in nickelalloys by reheat treatment, J. Jnst. met. 1968 13-4

Type of defect Type of loading Repair procedure Problems

- Fatigue cracking LCF Removal of arlas possibly inciplently Complete removal in bores, holes cracked by machining (regeneration] gears - Fretting in bolt HCF, LCF Shot peening, removal Paremeter monitoring, holes and crack detection grooves - Worn contact Wear Metal spraying Adhesion surfaces - Abraded seal Wear, Weld build-up, micro-plasma Welding cracks, defects lips thermal fatigue spraying - Worn grooves Wear, HC' Electroplated protection against wear HCF strength, _ I ______process monitoring

Table 1 Repair Procedures for Shafts. Rings and Discs

Type of defect Type of loading nepair procedure Problems

-. Coating damage Thermal fatigue, lVechanical, chemical stripping Residues, wall thickness erosion, corrosion

- Creep damage Creep-rupture Regeneration Effects on material, extreme damage to certain components - Abraded seal Wear, High-temperature brazing, Welding defects lips thermal fatigue weld build up

- Thermal Thermal fatigue Reduction of oxides, high-temperature Cleaning of small cracks, fatigue cracks brazing effects on material "-Notches FOD Blending Smearing of starting cracks

Table 2 Repair Procedures for Turbine elades and V..c. V rD r0'3

Figure 1 Detail of a Technical Repair Instruction for the replacement of ahraded sealing fins by a high-temperature brazed piece of suitable material, tinal shaping by grinding

Figure 2 Microsection showing repaired sealing fins, blade material IN 100, brazing filler material Ren6 80 + 2 % B

Figure 3 Welu build up of worn fins, TIG welding, filler material Nimonic 90. blade material IN 100

2 igwre 4 Weld build up of abraded blade tip i-,

i * s.- -

,, °

Figure 5 Repair of vane material by soaking in reducing atmosphere and high temperature brazing

4."

a... •-.*. ,.

• .. sa Ic o.WfHC ýWO~

Figure 6 Consumption of creep life of sets of turbine blades vs. service time, as assessed by taking the mean creep damage of a test sample from each set

AIWt ! ---

nr~ \•. rat W

Figure 7 Creep rate of specimens with and without regeneration treatment J , ,-

4:* V •'

L- "n~tured te sec r bOjetr- Hip C ter HtD lOS • iit,' -

Figure 8 Grain boundary crac~kinig in a creep rupture tested specimnen before and

Before HIP (1Ox) After HIP (lOOx)

Figure 9 Microsections of blades before and after HIP treatment

90 + regeneratEd

700

100 Iruptire

Figure 10 Rupture probability of blades tested in hot qas atmosphere, with and without regeneration treatment 14-1

A NEW APPROACH 1C TiHiEWELDARILI K ;IC.LAZZ,

A:Y-CAPT 1r PL"ER M.E;TATI gRY .- P-EFA/LTY.

ELBAR K-V. industrieturren Spikweien 5%3 AC LOMM The Netherlands S~SUgN'ARY

The investigation has shown that crack for"ation in the HAZ associated with the wel'inm of high r' -content nickelb-se superalloys is due !.ainly to the shri nka.-, which takes place during precipitation of the Kr . Pr •, •-oation can be bye.ntntroQn the cooling rate during weldin' of these alloys. On the basis of measurements, which illustrate Lhe relatinnshipp.n wt.,en n -0an cool's,- rate after welding, the partial effects of the elements Co, Cr, Al nd Vu on thn weidability were calculated, and this allowed a modified weldabiity dialram for nickelbase superalloys to be established. Simple hardness measurements ippear to &ve really useful results for crack-free TIG welding, plasma wellinq, friction welding and EB welding of nickelbaoe superalloys. Small diffe'aenoes in chemical composition and the degree of homogeneity -f ' structure can hav a iv efect en the welding behaviour of superalloys.

1. INTRODUCTION The repair and recovery of both the stati-nery ani the rotat'in-n parts of the hot section of a gas turbine is becoming more and more a 7pnerall.-aacepte! practior. The regularly repeated assertion that repair of the hot section, and in particular the first stage inlet guide vane and rotatinq blades and buckets is not possible, in based more on commercial considerations than on any practical and/or theoretical basio. In the special case discussed in this report, concerning the repair of nicke'base superalloys, such as are used in the first 2nd stages of the rotatinn sections of a gas turbine, welding plays an important -art. Zome typos of dama'e, e.n. tip weer and FOD can be repaired economically only by welding. The application r7 wel-in-" for the repair of rot.ting parts, and in particular blades end buckets, is nonetheless affected by a number of factors, namely : a) the stress level in the region of the blade, which is to Oelled. b) the temperature level in the region cf the blade to be repaired. c) the wall thickness of the region to be welded. d) the material of the part to be welded. As the wall thickness of the rotating parts to be repaired increases, accompanied normally by a higher level of stress, the weldability of the hirhl'y-stable nickelbase superalloys used for these parts falls rapidly. The cobaltbase superallo:s are generally characterized by muon lower strenoth, but this is accompanied by much better weldability. Their use is generally confined to the stationary parts of hot section. A fortunate circumotunnc- fro. the rep ir reints ý view is that an increase in the stress level from the tip to the root of a blade is accompanied by a correspcndin ldecrease in operating temperature. The use of, for example, TI0 repair welds has bee'n hitherto mainly restricted by the following factors 1) a Jack of commercially available nickelbase superalloy filler wires of oatisfactery high-temperature strength (creep characteristics) suitable for welding the hirhly stressed regions of blades. 2) the decreasing weldability as the wall thickness increases in the case of '-strengthened nickelbase alloys as used for the highly stressed regions of blades. These two factors more or less balance each other. Up till now very little research has heen carried out into the factors, which affect the w-ldabilAty of high V -content nickelbase superalloys. It is moreover a fact that alloys, such as -n 10, Tn 93a and U 720 were developed not with a view to their weldiro characrenrstics, but on the basis on high temperature properties, such as thermal stability, ,r-ep ntrrngth and h.t corrosion resistance. 2. THE WEDING OF NICKELBASE SUPERALLOYS

One of the earliest publications in the field of the weldability of nickelbase superalloys dates from 1966 and provides a recognied model -f 'AZ cracking (Ref .) The greatest increase in practical repair work by hibar, partcuolarly in the ease of stationary turbine hot section components, began in the last five ,ear5. Due to the necessity for repairs to an extremely difficult to weld s'n-"rll'-", -h ain In 7/" f- example, the need arose for fundamental investi .ationn -f tho :I trrn, which affeC - lh weldability of this alloy. The most lmportant def-t s, which can -- 'ur when w-Ki' nickelbase superalloy3, which contain are HAZ crickinq and FW'T"-rikin". The investigation began 4 years ago under contract Fror the M inistro of I-rn'e- AC ,7,r 14-2

wxtlhi"i t.,!±t p . .. E~ef,7.!ah-e:ec n ." • k ". -":- : Z:.-'.• il:" •. "-:-,i.. I : .. -. . . A•W.* '•:

.v ; . .*'. .2 ; _ ; ' . k7 - '," .U:> . . • ... - - . -. t .. -. : - .

t U I•"r;-.-

$ ; t 4-5-- -,,.L .w.•. "aC : :÷ • .': : = - ? - - " :: -:,b - !- '.

nt- s l. A - Z.s 7n. r '-

s~nth~ rs-ea'.nz: .. i e -••h -T'-i-"=÷ <'t .nT1 h '- "'--:-: - h.e - :.i-.r:l t2CX58 S'dOV.}e:;-. ,. -.- * oh. 2s he rea I- n-i.: - -w

- i th- ,n .s'z I' r r. o-I.' t i- :1 ,= :- -h -IcIniu 1

w l In'e uS C.: '2"7-

"in100--S- >4 z, -

,'T.s,'-" S•- ~'. h ~ ec 'itl- - " - ' r eI' s-a -it :"mvnc f th~s e .1.-uoh>'. the •;

i • C~~~~~~~~T5-7 •(3"• : {"•?? s.-

ti -n 752

"(a -Itint 'ao it . -:,Tlý,;h'e

iC t1n t12 - - it i s " -h--

X Z ": ","

1 t clat' that t.-Ž e Znt n - f- D ".n

R -ANY •-- 14-3 in thu case 3f in 7AS the f'g ,,wfri "pr-nipj,, tion-rat- nu:- -r:" w,:, fut

0050) WK) (.N;84) 7(r

Al T!i Cr

Co is more or less indifferent, C ins astron4 inhnbto-r of4 -mipitrti.n, Ti is a stronger• r -forr:r than AL!, (in atmie py'r.,rt'o ,

Note : Atomic percentages only are usedc or -sa-e calculations. Thus, H hinhr T-'A- ratio can result in a higher rate of '-precipitatin withur any; effect r the Av number.

When we graphically plot the compositions of" a rumber of well-known sueralloys, determined from empirically-derived formulas, a much clevrer picture nf the effect of ti,. elments on the wecdability is ontaond ClAe fiure nc " ) This nicture differs t distinctly from graphs as.2pild by other in''est,-a ipr YrF 3 il ;).

3. APPLICATION OF THE h.EU.:.T O TYE . =TGAT- " 'TOWLDINGFROCK=0"• ..... NE. .. TIC- ;: A•." WELDING.

The differences in the electron beam welding b'ehavior (cr'ck susceptibility) between two apparently identical I 790 P.M. 1llvys can be satisfacterily explained vn the basis of a difference in precipita1tion behaviur. cn,-th- empirically dr'trmtin.: effects of composition on the we 4 lablity diagram, the nompositi-ns of two 7 T00 F.H. types (U 700 P.M. A and B!, whiah clearly "i ,.:d in the. err'' I plotted on the diagram. This showe, that the positivna Y theset ll,.y were nlo distinctly different. The more crack-sensitive U 709 PM. type B was appreciably dispaced tow'•rP a ro,>n of higher crack sensitivity. in view of the fact that the measured siza' of thc V-particles were not user in the determination Of the ,..:.li ".,.l- nn'--p !a "hhi case of the U 700 typos mentioned, this results can be considered. s ramnrkab. Measurements -ale later and which illustrate the relationship bu tw-n hardness, maximum cooling rate and i'-particle size are presented in Fizurn 5 and 9. The m re crack sensitive type B appears to produce a lower hardness fr the same 'vixcoum coolln, rate. On the basis of these results the welding parameters were modified in such a manner that maximum hardness was produced in the -AA, resultimn in a crack free weld. It seems therefore that there are two parameters which rice ini itoa of the wcldinc, behaviour of high V' -content nickelbase superallcys. These are a) the compositional balance of Co, Cr, Al and Ti. b) the hardness measured in relation to the cooling rate. These two factors form in fact the basis for the development of - weldlnq method with less crack-sensitive, '(-containing nickelbase superalloy filler wires with sufficient creep strength to allow welding of the highly-stressed regions of blades and buckets. The results of the investigation appear applicable to the friction weldine of superalloys, but no further observations as regards this can be made here. Since preheatino before welding obviously hUs an evl'eiit on the saximum cooln,j a.tvtit w-u' ,.. thut crack-free welds can be produced by this means. Wnr niekelbase superal)oyz, however, this is not really effective until the temperatures are above R0OW. Figure 10 shows the effects of several preheat temperatures on the hardness pattern in the HAZ of in 73F. EB welding of nickelbase superalloys with preheat has been recently reported in the literature (ref, 2) without any rigorous theoretical explanatinn, Pruheatni- for EB welding will, however, involve temperatures above 900"C. An important conclusion, which also can be made is that pre-weld h,at trea'-,nt can have an effect on the weldability of nickelbase superalloys only if it has a di effect on the homogeneity of the I - "r -structure. It ;Y Prl- K"-ý1 reason that r weld hip treatment can sometimes have a positive effect on the weldability of th 1 'eult to weld superalloys, 14-4

•.:•: i~i-FFER N-:CE 177':

1. W.A. Ow-azarski, D[.. Duall and C.P. Sullivan. A model f1•• Heat -af fe cted zone r ranking in Hi-bqsve superal-loys. Welding Journal H.S. 19F6, !495-157

P. EB H:KUiiin ,f P.M. ii-tws.. :;sqprailyn. H. Plhainbul'uy, W. ia 11 ace %Ut. A/.pen:,atkal 1'h;t/1 ]',;hiIfrnt Ottawa rnwint Paper proL•hen'd at a P.'M. sutperl• loyz eoof.rerne in .ovboer' 1180 in Zurich, SwitzerlanO.

* !Is1113 Ii '.Ž The- tr-,r n J..:hr) WIley and -)ons, Nelow Y,\rk 1971 SPBN n-0471-79207-1

-. Owezarski

Process and Metllurli.i]. factors in Joining uperallo:yn ind other high Service temperature materials. AGAHD report - LS-91 * - 14-5

HARDNLSS SURVEY ACROSS TH'E IA4 OF INCONEL 738 AS A FUNCTION Or VICXERS THE COOLING RATE AFTER WELDING HARDNESS

650

360

TEMP. CYCLUS NO. 2 A T .

400

ams

STo .mCCL..N.."'. IE

DISTANCEWEL INllMFRO THE...... 14-6

.'l'.AN WIPAwRTiz'LE A

*-~~-~**~ ,AM"A N !TMTGL Dl&VFI'MIF•'RS AS A FUNCTION OF THE MAXIMUIM COOLING RATE.

2000

4- A.

4r 4

ci' I

11gILI. DO O22RCIIA NA

i1 ~ ~ ~ ~ ~ ~ ~ 'FI~APRX 1 2l IIN,'C. ORAPR

CK MAXIMUM COCLINQ RATIE

T114E: IN SrCONDS Fiq. 3 14-7

VICKERS HARDNESS OF INCONEL 738 AS A FUNCTION OF GAMMA PRIME PARTICLE SIZE

* MEASURED AVERAGE OF SIX OR MORE MEASUREMENTS.

480-

4604

430-

Boo SLO 'coo 1250 -400 3000

GAMMA PRIME PARTICLE SIZE IN

Fig. no. 4

WHOM 0 04

z-ID

-6

4,0 0 0 0 -•• ~*~4*oi O•

H -Q r'

00 4 U 0.

CIz

E-4 z 10 i iU

Fig no 14-9

0 C o -•

m()

o* 0J 0

Fig. no. 6 14-10

10-

Si iinI 10 DIFFICULT TO/ / /i- WELD /

S/ aMerl 78 HAZ CRACKING BUT

SAWELDalE AT LOW WEIDINGSPEED / /-

/ *U700* 3m B _7- U70 P-m A eB190M TU710 PWHT CRACKING i / / tlr M eO0 -/ FAIR WELDABILITY SI el,738

/ I-*UUCO ,~~~o i•I / / U5001

iI -- • B- I I, tn 9-32t,

/ fln 700 / -Um-READILY So~eni i 41- WELDABLE •i•, •,• 4- / / / *Rmnsi~l

DECREASED P8 S-/ •WELDX'BILITY - * BA0 I -

INdREAS /

AGIN DECREASED . RESPONS 1 AGING -

/. ESPONSC

*ib/

• ,, o •• 6 4 '- 7 .. •W96 -g!!7 0.28Cr+O.043Cro]

-• Illrig. no. 7 14-|1

o oo I--

.4•

I DI

4. w t-4 i'i| 0 -m ° IXU

mR H 0 • i-= II ug

al -• I! !! !-

/'.14..Il. LLI

/ I II- Fig.III 14-12

o Q

nI II . 4 cu

al 0

to / Iii

\ \a

0 Ea 110 0

Fgo9

0 •

tjl

I I I I I I" I

S• Fig. no. 9 14-13

VICKERS HARDNESS SURVEY ACROSS THE HAZ OF A FUNCTION OF THE HARZDNESS PREHEATINCONEL TEMPERATURE738 AS

0 5503 -- *0

500

Soo 0

.- 0 0 o0 0 TC 3

0 C

350

TEMP. CYCLE PREHEAT TEMP.

O TEMP. CYCLUS NO. 3 : 20"C TEMP. CYCLUS NO. 4 : 300,C O TEMP. CYCLUS NO. 5 : 6001C

aO 0!,t 2.8 ! 1s 2 '0 2B

DISTANCE FROM THE WELD IN MM

Fiq. no. 10 R3-1

RECORD'S REPORT - SESSION IlI

Dr James A.Snide Director. Materials Engineering University of Dayton College Park 300 Dayton. O1t 45469, USA

The first two papers of this session continued from the pievious discussion of hot isostatic prLsing (1111) to rejuvenate cast nickel base turbine blades. The third paper considered the weld repair of as-cast and powder metallurgy superalloys.

IThe first two papers showed the promising benefits due to HIP processing as a regenerative treatment to extend the life of gas turbines components removed from service during overhaul. The focus of these efforts were on the removal of creep damage during 11lP processing. The importance of controlling the cooling rate after the 11iP operation was emphasized. The following questions were discussed but not resolved.

(I) How many times can a turbine blade be rejuvenated?

(2) What is the effect of HIP rejuvenated on properties other than creep. that is. microstructure. fatigue behavior. etc.

The paper on weld repair of as-cast and powder metallurgy presented a framework for evaluating heat-affected-zone cracking as a function of composition and welding speed. The paper presented an interesting way to look at a very complex problem. It appeared that the data might be treated in a more fundamental way using thermodynamic. nucleation and growth and diffusion theory. AAi

APPENDIX

COMMENTS ON THE MAINTENANCE IN SERVICE OF HIGH TEMPERATURE COMPONENTS IN AIRCRAFT JET ENGINES

George C.Deutsch 8303 Whitman Drive Bethesda, MD 20817. USA

I first became acquainted with jet engines in 1946 and since that time I have. of course. seen many changes. Today'., engines are far more powerful and fuel efficient : but they have lost one of their most beguiiing features %inplicity. Today's engines with their multiple stage turbines and concentric sW'f~s are very complex and therefore much more difficult to maintain.

In 1946 military engincs had a time between overhail of 150 hours but very few of them attained even this time interval. The blade temperature was about 725(C and it has grown at the rate of about I 5C per year cvcr since. A Irul1 remarkabk achievement when all during this period metallurgists kept predicting that they had pushed alloys about as far as they could go.

It is perhaps mort clear today that at any time in !he past that the end is not in sight and I believe we should not only be working on extending and predicting the life of current engine components but we should devote considerable attention to making higher temperatures possible. It should. of course. be noted that in the past we wanted higher temperatures to enable aircraft to fly at higher speeds. Today we want these same higher temperaturcs so that the engine operates in a inorc fuel efficient manner. Some of the studies that I believe should be undertaken to prepare us for the future are listed below:

(a) We should seek a far mnore precise definition of the environment f(temnperatutre. srsand chiemical) than is available today. This definition would give us better clues to the mechanisms of failure and would enable the designers to come up with less demanding designs.

ib• We should develop far superior methods of keeping track of component histories so that we can better decide when rejuxenation procedures are appropriate.

(c) Surface crack cleaning still seems to me to be a major problemn and better methods are needed before we can be confident of the repairs.

(d) I think therv is further room for improvement in the presently used braze alloys. WVeare currently adding boron or other elements to reduce the brazing temperatures, however, it would appear to me that if wc could find bubstitutes which would completely vaporize during the brazing processes, superior repairs could be made.

te) New techniques are needed in both NDF. (particularly for coatings and cooling passages) and welding. In the latter case I believe lasers should be looked at intensively.

(f[ We should be getting ready for the newer materials that are already finding their way into jet engines. The principal one it, of course, conventional alloys made by powder metallurgical niettotls. single crystal blades. inter-metallic compounds and ceramics will gradually follow along. We should be prepared to inspect and maintain these new materials.

I would like to close with one final comment. We should also be very conscious of the gradually deteriorating quality of jet engine fuels. We already have seen a gradual increase in the percentage of aromatic,%these fuels contain and we know these can cause severe erosion problems and temperature increcases in the engipe. Other changes in fuel chemistry can introdutce vurrosivc ing-redients. We must ever be mindful of these chang-s in planning our maintenance procedures for jet engines. REPORT DOCUMENTATION PAGE ill[-I.Recipient's~ttfemeuce 2.Orignator's Refere S1• 3. FnetetteFurdker Reference .Seeurity tlauiiicstioa of Document I AGARD-CP-317 ISBN 92-835-0307-4 UNCLASSIFIED

5.Originator Advisory Group for Aerospace Research and Development North Atlantic Treaty Organization 7 rue Ancelle, 92200 Neuilly sur Seine, France ""~~ -- .Title

MAINTENANCE IN SERVICE OF HIGH TEMPERATURE PARTS III. 7.P. ...resen . ted. at ...... t at the 53rd Meeting of the AGARD Structures and Materials Panel held in Noordwijkerhout, the Netherlands 27 September 2 October 1981. 8.Author(s)/Editor(s) 9. Date Various January 1982

10. Author's/Editor's Address I1.Pages

Various 172

12. Distribution Statement This document is distributed in accordance with AGARD policies and regulations, which are outlined on the Outside Back Covers of all AGARD publications. 13=KeywordslDesriptors

Engines Maintenance Components Fatigue life High temperature tests Service life

14.Abstract

All NATO countries need to combat the increasing cost of maintenance of engines and scarcity of strategic materials by improving component utilization. The objective of this Meeting was to review the problem areas and experiences in the maintenance of high temperature parts; many of these problem areas having a common base in relation to service experience and the characteristics of material behaviour, so that users may benefit from the advances in materials science and the future needs for R&D may be identified.

/i I

...... j., r-

F~ °--. r

ezi 2 z = ,•- • _ • • o-o_"•• • I- • H --

<-; 6-- 6- _. . .0 _ - S- >.

, C .001Vj * 6 6--

CL Co

r- -

r .r

2 .- L ia

- .~.> .t-Q 3 ý 'I ' • • -- '- u :z = = - • c.. E - - _

3) - - 7 v .0 1.

< It.z;in'P

Iz-j 1;> • L, c,••• _= 'C< , • " ,. . . .2 .6 - - -

iz o E o.-

i~~~' m